Xi3 Predictive Analysis User Guide en

BusinessObjects Predictive Analysis User Guide
BusinessObjects Predictive Analysis 3.0 Windows
Copyright
2008 Business Objects. All rights reserved. Business Objects owns the following U.S. patents, which may cover products that are offered and licensed by Business Objects: 5,555,403; 5,857,205; 6,289,352; 6,247,008; 6,490,593; 6,578,027; 6,831,668; 6,768,986; 6,772,409; 6,882,998; 7,139,766; 7,299,419; 7,194,465; 7,222,130; 7,181,440 and 7,181,435. Business Objects and the Business Objects logo, BusinessObjects, Business Objects Crystal Vision, Business Process On Demand, BusinessQuery, Crystal Analysis, Crystal Applications, Crystal Decisions, Crystal Enterprise, Crystal Insider, Crystal Reports, Desktop Intelligence, Inxight, the Inxight Logo, LinguistX, Star Tree, Table Lens, ThingFinder, Timewall, Let there be light, Metify, NSite, Rapid Marts, RapidMarts, the Spectrum Design, Web Intelligence, Workmail and Xcelsius are trademarks or registered trademarks in the United States and/or other countries of Business Objects and/or affiliated companies. All other names mentioned herein may be trademarks of their respective owners. Business Objects products in this release may contain redistributions of software licensed from third-party contributors. Some of these individual components may also be available under alternative licenses. A partial listing of third-party contributors that have requested or permitted acknowledgments, as well as required notices, can be found at: http://www.businessobjects.com/thirdparty
Third-party Contributors
Contents
Chapter 1 Predictive Analysis 7 Chapter 2 Predictive Analysis populations 9
Adding a population in Predictive Analysis................................................10 Editing a population in Predictive Analysis................................................11 Removing a population in Predictive Analysis...........................................12 The Populations list in Predictive Analysis................................................12 Chapter 3 Derived Variables in Predictive Analysis 13
Derived variable as fact table aggregates.................................................14 Creating derived variables in Predictive Analysis......................................15 Creating a binning in Predictive Analysis..................................................17 Editing derived variables and binning in Predictive Analysis.....................18 Deleting derived variables and binning in Predictive Analysis...................18 Chapter 4 Predictive Analysis models 19
Adding a model in Predictive Analysis.......................................................20 Predictive Analysis model data types...................................................21 Copying a model in Predictive Analysis.....................................................22 Editing a model in Predictive Analysis.......................................................23 Deleting a model in Predictive Analysis.....................................................23 Regenerating a model in Predictive Analysis............................................23 Regenerating statistics in Predictive Analysis...........................................23
Contents
Chapter 5
Predictive Analysis model-based metrics
25
Adding a model-based metric in Predictive Analysis.................................26 Aggregation functions for model-based metrics...................................28 Goal Level............................................................................................30 Deleting model-based metrics in Predictive Analysis................................30 Refreshing model-based metrics in Predictive Analysis............................30 Purge a model-based metric in Predictive Analysis..................................31 Publishing model-based metrics in Predictive Analysis.............................31 Goal & Metric Publishing Wizard..........................................................32 The model-based metrics list in Predictive Analysis..................................38 Chapter 6 Predictive analytics 39
Selecting a predictive model to analyze in a Predictive analytic...............40 Predictive analytic samples in the Analytics Catalog.................................40 Goal-based Influencer Detail analytic based on predictive models...........41 Configuring the Goal-based Influencer Detail analytic.........................42 Setting the display options in Goal-based Influencer Detail and Influencer Detail analytics.....................................................................................42 Individual List based on predictive models................................................43 Configuring the Individual List analytic.................................................44 Influencer Detail analytic based on predictive models...............................44 Configuring the Influencer Detail analytic.............................................45 Setting the display options in Goal-based Influencer Detail and Influencer Detail analytics.....................................................................................46 Influencer Gains Chart analytic based on predictive models.....................46 Configuring the Influencer Gains Chart analytic...................................47 Key Influencers analytic based on predictive models................................48 Configuring the Key Influencers analytic..............................................50 Metric Forecaster analytic based on metrics.............................................50 Configuring the Metric Forecaster analytic...........................................51
Contents
What is a metric forecaster detrended mean?...................................101 Model Gains Chart analytic based on predictive models.........................102 Configuring the Model Gains Chart analytic.......................................103 Model Gains Chart tips.......................................................................103 Variable Profile Box Plot based on predictive models.............................105 Configuring the Variable Profile Box Plot analytic..............................106 Predictive Analysis Terms........................................................................107 What is binning?.................................................................................107 What is a Boolean?............................................................................107 What is a derived variable?................................................................107 What is a goal?...................................................................................108 What is goal status?...........................................................................109 What is goal-based binning?..............................................................109 What is a grain?.................................................................................110 What is an influencer variable?..........................................................110 What is a joiner?.................................................................................112 What is a leaver?................................................................................112 What is a metric?................................................................................112 What is a population?.........................................................................112 What is a predictive model?...............................................................113 What is the root-mean-squared-error?...............................................113 What is the rules engine?...................................................................113 What is a sampling?...........................................................................113 What is a variable?.............................................................................114 Chapter 7 Predictive Analysis Terms 115
What is binning?......................................................................................116 What is a Boolean?..................................................................................116 What is a derived variable?.....................................................................116 What is a goal?........................................................................................117 What is goal status?................................................................................118
Contents
What is goal-based binning?...................................................................118 What is a grain?.......................................................................................119 What is an influencer variable?...............................................................119 Predictive Analysis model data types.................................................119 What is a joiner?......................................................................................121 What is a leaver?.....................................................................................121 What is a metric?.....................................................................................121 What is a population?..............................................................................121 What is a predictive model?.....................................................................122 What is the root-mean-squared-error?....................................................122 What is the rules engine?........................................................................122 What is a sampling?................................................................................122 What is a variable?..................................................................................123 Appendix A Get More Help 125
Index
129
Predictive Analysis
Predictive Analysis
BusinessObjects Predictive Analysis allows you to build models that show the relationship between variables and determine the key influencers of that specific or goal variable. In Predictive Analysis, you work with: Populations You identify the groups that you want to analyze. This is not necessary if you are creating a Metric forecaster analytic. Derived Variables You can customize variables that refine your results and do binning of the variable contents to show trends by groups. This is not necessary if you are creating a Metric forecaster analytic. Models You configure a model analysis that includes populations, derived variables, influencers, and goals that you can put into an analytic. This is not necessary if you are creating a Metric forecaster analytic. Metrics You can create model-based metrics that you use to improve your models. Predictive Analytics Predictive Analysis provides you with pre-defined analytics that you can use with your models.
Predictive Analysis populations
Predictive Analysis populations Adding a population in Predictive Analysis
Any analysis requires the definition of the data to be analyzed. The data for the following influencer analytics is defined in a population: Influencer Detail Key Influencers Influencer Gains Chart Model Gains Chart Variable Profile Box Plot Individual List
This section shows you how to work with populations. You can do the following: Add, edit and remove populations Change the population list view
Related Topics
Influencer Detail analytic based on predictive models on page 44 Key Influencers analytic based on predictive models on page 48 Influencer Gains Chart analytic based on predictive models on page 46 Model Gains Chart analytic based on predictive models on page 102 Variable Profile Box Plot based on predictive models on page 105 Individual List based on predictive models on page 43
Adding a population in Predictive Analysis

1. In Predictive Analysis > Populations, select from the drop-down list a population area in which you want to create your population. 2. Click Add. The Create Population wizard opens. 3. In the Set step, select a set type from the list.
Note: You can view the source set list by sets or groups. To create a set,
see the Set Analysis documentation. If you select Single Set, you need to select a subset. If you select Migrants, Overlapping Members, Same Time Joiners or Same Time Leavers, you need to select a destination set.
10
Predictive Analysis populations Editing a population in Predictive Analysis
4. Specify the time period: To specify a period, activate During last and set the number of periods. To specify a month, activate For and select the month and year. 5. Click Next. 6. In the Filter step, select conditions in the Available condition list and use the double arrow button to move them to the Selected conditions list. 7. Click Next. 8. In the Attributes step, type the name for the population you are creating. 9. If you want others to be able to use this population, activate Public. 10. Click Finish. The population you created appears in the Populations list. The right-side panel shows the details of your population.
Related Topics
What is a joiner? on page 121 What is a leaver? on page 121 What is a population? on page 121
Editing a population in Predictive Analysis

1. In Predictive Analysis > Populations, select from the drop-down list a population area in which you want to create a population. 2. Select the population you want to edit. 3. Click Edit. 4. The Edit Population wizard that appears follows the same workflow as that for adding a population.
Related Topics
Adding a population in Predictive Analysis on page 10
11
Predictive Analysis populations Removing a population in Predictive Analysis
Removing a population in Predictive Analysis

1. In Predictive Analysis > Populations, select from the drop-down list a population area in which you want to create a population. 2. Select the population you want to remove. 3. Click Remove, then OK.
The Populations list in Predictive Analysis

In Predictive Analysis > Populations, you can view the Populations list in the following ways: Flat list Select this option to see a flat list of the populations for a Population area. By Sets Select this option to see the list of populations by sets.
Note: This option is only available if the population is based on a set
universe. Show Groups Select this option to see the list by set groups. By subsets Select this option to see the list by subsets, for example by joiner or leaver subsets.
12
Derived Variables in Predictive Analysis
Derived Variables in Predictive Analysis Derived variable as fact table aggregates
When you are setting up an analysis, you can customize a variable to extract specialized information. For example, from the variable 'Owner' you can extract the owners in a certain age range. Instead of creating a variable in the universe, you can create a derived variable in Predictive Analysis. In addition you can refine the results for 'Owner' by binning the results by age. You can use measures with arithmetic or logical operators to form more complex expressions. With derived variables, you can: specify different date restrictions on measures with calculated aggregations from fact tables. create duplicates of universe measures for cross-sell analysis. create a variable that represents membership in a specified set (for example, member of "Frequent Buyers"). experiment with private variables representing alternative approaches before choosing one to make available to others.
You must create a boolean goal variable if you want to use the following analytics: Influencer Gains Chart Model Gains Chart
In Predictive Analysis > Derived Variables, you can create, edit and remove derived variables and binninsg.
Related Topics
Influencer Gains Chart analytic based on predictive models on page 46 Model Gains Chart analytic based on predictive models on page 102
Derived variable as fact table aggregates

Predictive Analysis allows you to define derived variables using a formula editor. Derived variables allow users to use fact table data in Predictive Analysis.
14
Derived Variables in Predictive Analysis Creating derived variables in Predictive Analysis
Derived Variable Syntax for Measures
The general syntax for measures used in derived variable definition is as follows (square brackets [] represent optional parameters):
<m:Measure name,Null value replacement constant,[<t:Aggregate type>, [Date expression], [Date expression]]>
For example:
<m:Profit, 0,<t:Sum>, {1/1/2002}, CurrentDate()> Sum of profit from 1/1/2002 until today, with nulls replaced by 0 <m:Order Size, 100,<t:Maximum>, CurrentDate() - 30*6, Current Date()> Maximum order size in the last 180 days, with nulls replaced by 100 <m:Order Size, 0, <t:Maximum>, Date(CurrentYear(), CurrentMonth()-6, CurrentDay()),Current Date()> Maximum order size in the last 6 months with nulls replaced by 0
Creating derived variables in Predictive Analysis

1. In Predictive Analysis > Derived Variables, select a variable subject area from the dropdown list. 2. Click Add. 3. Select New Derived Variable. 4. In the Create Variable panel, type a name for the variable in the Variable name box. 5. If you want others to use the variable, activate Public. 6. Select the Data type: Boolean Date Numeric Character
Note: Formulas using a date can be treated as a data or numeric value,
representing a monotonically increasing number of days. When used in a model definition, avoid explicit dates such as 'date of birth'. Use instead numbers representing differences between dates, such as 'age' defined, for example, as currentDate() - DateOfBirth.
15
Derived Variables in Predictive Analysis Creating derived variables in Predictive Analysis
7. Specify variable derivation in the Formula box by double-clicking items from the Objects and Functions and Operators lists. You can use the following source data: Dimensions from the universe corresponding to the selected subject area Measures from the universe Membership in a set associated with the selected subject area, if the subject area is set-based Date functions that return the current date or current month, for example, based on the application server clock Combine or transform date functions using operators, functions and aggregations to arrive at a derived variable definition. There are two types of measures from the standpoint of derived variable definition: measures with prompts and measures without prompts. Measures without prompts are used 'as is.' An example is a measure representing a numeric field of a dimension table, like 'age.' The other kind of measure is a time-based aggregate, such as 'sum of revenue in the past 30 days' or 'time (sum of days) since last order.' In these cases, additional information is required to resolve the value: Null value substitution constant Aggregation type Starting date Ending date
Note: After inserting a measure that represents a time-based aggregate
into the formula area, enter appropriate values for the parameters just before the closing angle bracket (">") for the measure definition in the formula area. To use membership in a population for a derived variables definition, you must use the isinpopulation() function with the Population Name parameter. 8. To add a specific day to the formula, click the icon next to Dates. 9. Click Save.
16
Derived Variables in Predictive Analysis Creating a binning in Predictive Analysis
Creating a binning in Predictive Analysis

Binning is a way of compressing all values for a variable into a smaller number by subdividing the range of possible values into groupings or "bins". 1. In Predictive Analysis > Derived Variables, select a variable subject area from the dropdown list. 2. Select the variable on which you want to create the binning. 3. Click Add. 4. Select New Binning. 5. In the Create Binning panel, type a name for the variable in the Binning name box. By default the variable name is used, however it is best to use a unique name. 6. To allow others to use this variable activate Public. 7. Select the Variable type. 8. Set the number of bins. 9. Select a Calculation option: If you are binning based on numeric variables, the options are: Equal count for each bin The option creates bins in which the number of records in the specified working population that fall into each bin is approximately the same ("binning into quartiles"). Equal width for each bin The option creates bins whose value range is the same width, for example, a binning with five year intervals. If you are binning based on Nominal variable, the options are: Equal count for each bin The option is the same as that for numeric variables, except that there is no assumed ordering in which values are grouped. Equal number of categories The option arranges for each bin to group the same number of distinct values. For example, a binning for State of Residence can have each bin contain the same number of states. For Nominal variable, choose a small number of bins, for example, two, and then create additional bins via Edit.
17
Derived Variables in Predictive Analysis Editing derived variables and binning in Predictive Analysis
10. Select the population that you want to use to define the initial bin definitions. The reference population provides quick feedback on the number of records that fall into specific bins. After the bin definitions are saved to a binning, the population is no longer retained.
Note: You can change the view of the Available Population list.
11. Click OK.

Related Topics
What is binning? on page 116 The Populations list in Predictive Analysis on page 12
Editing derived variables and binning in Predictive Analysis

1. In Predictive Analysis > Derived Variables, select a variable subject area from the dropdown list. 2. Select the item you want to edit. 3. Click Edit. If you are editing a variable, the Edit Variable panel matches the Create Variable panel and workflow. If you are editing a binning, the Edit Binning panel matches the Create Binning panel and workflow.
Related Topics
Creating derived variables in Predictive Analysis on page 15 Creating a binning in Predictive Analysis on page 17
Deleting derived variables and binning in Predictive Analysis

1. In Predictive Analysis > Derived Variables, select a variable subject area from the dropdown list. 2. Select the item you want to remove. 3. Click Remove, then OK to confirm.
18
Predictive Analysis models
Predictive Analysis models Adding a model in Predictive Analysis
Once you have defined the population and optionally any derived variables, you need to build a model.
Note: To create a model, you must have a population; however derived
variables are not required.

Related Topics
Adding a population in Predictive Analysis on page 10 Creating derived variables in Predictive Analysis on page 15
Adding a model in Predictive Analysis

1. In Predictive Analysis > Models, select a model subject area from the dropdown list. 2. Click Add. The Create Model wizard panel appears. 3. In the Population step, select a population from the Available Populations list.
Note: You can change the view of the Available Population list.
4. Click Next. 5. In the Influencers step, select objects by doing one of the following: Double-click an object. Highlight an object and click the double-arrow button. When an object moves to the Selected objects list, the data and binning types appear. 6. Select the data and binning types from the lists. 7. Click Next. 8. In the Goals step, select objects by doing one of the following: Double-click an object. Highlight an object and click the double-arrow button. 9. 10. 11. 12. Click Next. In the Attributes step, type a name for the new model. To make the model available for use by others, activate Public. Set the Refresh type:
20
Predictive Analysis models Adding a model in Predictive Analysis
Activate With Set to precipitate a model refresh whenever the set being used in the associated population definition is refreshed. Activate Independent to have a refresh that is scheduled independently of set refresh - for example, through Dashboard Builder Scheduler or a Performance Manager rule.
13. Set the Refresh scope: Activate Refresh Statistics only to associate the refresh of most statistics and metrics with the model whenever the model is refreshed. Activate Regenerate Model to allow the refresh the statistics, as well as the regeneration of the optimal binnings and computation the formulas used in generating model scores. This option is used usually with the Individual List analytic. A few statistics are also calculated only at the time of model regeneration - Net Relevance, in particular.
Note: Regenerate Model requires much more computation time than
Refresh Statistics only, because it requires multiple passes through the data. 14. Click Finish.
Related Topics
Predictive Analysis model data types on page 119 The Populations list in Predictive Analysis on page 12 What is binning? on page 116
Predictive Analysis model data types

Objects influence how the data appears in the model.
Sources Desktop Intelligence Objects Numeric Date Source data type Influence data type Nominal Goal data type
Character
N/A
Continuous, OrdiContinuous, N/A nal N/A N/A
21
Predictive Analysis models Copying a model in Predictive Analysis
Sources Derived Variable Character Numeric Date
Source data type Boolean Nominal
Influence data type Nominal N/A
Goal data type Boolean
Continuous data type
The continuous data type is the default data type corresponding to numeric variables. Do not use this type for variables that represent numeric codes rather than actual numbers (for example, zip code); treat such variables as nominal variables.
Nominal data type
A nominal data type has a value that is not inherently ordered, for example gender. All character objects from a BusinessObjects universe are treated as nominal.
Ordinal data type
Ordinal variables are ordered, but lack proportionality, as in the example of an undefined distance between adjacent values.
Related Topics
What is a Boolean? on page 116
Copying a model in Predictive Analysis

1. In Predictive Analysis > Models, select a model subject area from the dropdown list. 2. Select the model you want to copy. 3. Click Copy.
22
Predictive Analysis models Editing a model in Predictive Analysis
A copy of the model appears in the list.
Editing a model in Predictive Analysis

1. In Predictive Analysis > Models, select a model subject area from the dropdown list. 2. Select the model you want to edit. 3. Click Edit. The Edit Model panel matches the Create Model panel and workflow.
Related Topics
Adding a model in Predictive Analysis on page 20
Deleting a model in Predictive Analysis

1. In Predictive Analysis > Models, select a model subject area from the dropdown list. 2. Select the model you want to delete. 3. Click Remove and then OK to confirm the action.
Regenerating a model in Predictive Analysis

You need to regenerate the model if you changed variables and data. This updates the model structure. 1. In Predictive Analysis > Models, select a model subject area from the dropdown list. 2. Select the model you want to regenerate. 3. Click Regenerate Model.
Regenerating statistics in Predictive Analysis

You need to regenerate the model if you changed the data.
23
Predictive Analysis models Regenerating statistics in Predictive Analysis
1. In Predictive Analysis > Models, select a model subject area from the dropdown list. 2. Select the model whose statistics you want to regenerate. 3. Click Regenerate Statistics only.
24
Predictive Analysis model-based metrics
Predictive Analysis model-based metrics Adding a model-based metric in Predictive Analysis
Models create statistics, for example obsolescence. These statistics are exposed in Predictive Analysis as well as in time series analytics. Typically, a statistic calculated by a Predictive model quantifies a relationship between an influencer and a goal variable, but it can also be a descriptive statistic that summarizes some aspect of the influencer's distribution in the population with which the model is associated, for example, the mean of Age or the max of Purchase Size in Last 30 Days. Certain statistics can be valuable for automating model management, such as root mean squared error or goal variable obsolescence. These can be used in rules to detect when it is time to regenerate a model in statistics-only refresh mode or to trigger model regeneration when one whose refresh is unscheduled, for example, used for scoring lists, but not for statistics calculation.
Note: You only get values from the most recent Predictive model refresh.
In the Metrics page you can do the following actions on Predictive model-based metrics: Add
Note: In this documentation you learn how to create a Predictive
model-based metric, which is the required metric for an analysis. If you want to create generic metrics and manual entry metrics, refer to the Performance Manager documentation. Delete Refresh Purge Publish
Related Topics
What is the root-mean-squared-error? on page 122 What is a variable? on page 123 The model-based metrics list in Predictive Analysis on page 38
Adding a model-based metric in Predictive Analysis

1. In Predictive Analysis > Metrics, select a subject area from the dropdown list.
26 BusinessObjects Predictive Analysis User Guide
2. Select a set from the sets universe drop-down list.

Note: The option is available only for the subjects containing sets.
3. Click Add. 4. Select New Model-based Metric from the Add sub-menu, the Create Metric wizard appears. 5. In the Selection step, choose a model. 6. Optionally, select an aggregation function. The available functions depend on the model you select. 7. To select an influencer for a model, activate Select influencer. 8. To select a binning for the model, activate Select binning and choose a binning from the list.
Note: The option is not available if no bins have been defined under
derived variables and used in the model. 9. 10. 11. 12. Click Next. In the Attributes step, type a name for the metric. Type a description for the metric. Select the Calculation interval. The calculation interval provides the date condition for the metric. The duration of the time period that makes up the metric is called the metric's grain. Select a grain from the list. The list is created in the installation setup (see the Dashboard and Analytics Setup documentation for information on setting up the grain). 13. Select a Refresh type: Select With model so that the metric is refreshed when its model is refreshed. Select Independent to allow only manual refresh of the metric. 14. Select the Storage options: Select Store all metric values to use the entire hstory of the metric for trending purposes. Since metric values require little storage space, the option is recommended. Select Store last metric value only to store only the last metric value. Select Disable so that your model is not dependant on the last metric value. 15. In the Parameters section, select a default smooth option.
BusinessObjects Predictive Analysis User Guide 27
The Default Smoothing list allows you to specify the statistical transformation used in metric reporting when the Default Smoothing selection is activated in the Smoothing list. The nature of the data you are probing and which questions you hope to answer determine the type of smoothing that is most suitable. The administrator determines during installation and configuration the available transformations. 16. Select one of the following from the Trend is good when list: Increasing Select Increasing if an increasing trend is good. Decreasing Select Decreasing if a decreasing trend is good. On-Target Select On-Target if an on target trend is good. 17. Select an owner for the model. You can choose from any name that is in the Central Management Console. 18. Click Finish.
Aggregation functions for model-based metrics

Aggregation functions perform a calculation on a set of values to return a single value. The following tables illustrate the relationship between the data type and aggregation.
Influencer Level Importance Non-Numeric Data Type Measures usefulness of Influencer for "predicting" goal values If > 5, optimal binning can be obsolete (regeneration needed) Net Relevance Advanced metric calculated only at model regeneration time (Measures predictive power net of "stronger" Influencers) Measures predictive power net of "stronger" Influencers
Obsolescence
28
Influencer Level Data Quality
Non-Numeric Data Type %Missing, %Other, %OutOfRange
What about % Failed and Root Mean Squared Error?

Influencer Level Importance Numeric Data Type Measures usefulness of Influencer for "predicting" goal values If > 5, optimal binning can be obsolete (regeneration needed)
Obsolescence
Net Relevance (regeneration time Measures predictive power net of "stronger" only) Influencers Descriptive Statistics Data Quality Min, max, mean, variance, standard deviation %Missing, %OutOfRange
Influencer Bin-Level
Numeric Goals Bin Importance Signed Bin Importance Goal Divergence Weighted Goal Divergence Goal Rank
Binary Goals
Goal-Based - Influence
Bin Importance
Signed Goal Divergence Signed Bin Importance
Goal Rank Mean Goal Value Goal-Based - DistributionGoal Rate Mean Goal Ratio al Goal Variance and Stan- Goal Ratio dard Deviation
29
Predictive Analysis model-based metrics Deleting model-based metrics in Predictive Analysis
Influencer Bin-Level
Numeric Goals Count
Binary Goals Count Frequency (% distribution) Bin Mean (for numeric Influencers)
Non-Goal-Based - Distri- Frequency (% distribubutional tion) Mean (Influencer) Value
Goal Level
Alternative Model Quality Metric: Importance metric for the "prediction" or "score" variable Good choice for general-purpose model quality metric
Model Quality Metrics only Root Mean Squared Error (RMSE)
Obsolescence Detection Classification Rate (binary goals) R2 (continuous goals)
Deleting model-based metrics in Predictive Analysis

1. In Predictive Analysis > Metrics, select a metric group from the dropdown list. 2. Select from the list the metric you want to delete. 3. Click Remove, then OK to confirm.
Refreshing model-based metrics in Predictive Analysis

1. In Predictive Analysis > Metrics, select a metric group from the dropdown list.
30
Predictive Analysis model-based metrics Purge a model-based metric in Predictive Analysis
2. Select from the list the metric you want to refresh. 3. Click Refresh. A panel appears warning you that the operation can be long in execution. 4. Click OK to proceed.
Purge a model-based metric in Predictive Analysis

If you have refreshed a selected metric at least once, you can purge the metric history. You need to purge a metric when you want to empty the data in the metric history section and delete the metric data from the ci_probe_value table in the Performance Manager repository. 1. In Predictive Analysis > Metrics, select a metric group from the dropdown list. 2. Select from the list the metric you want to purge. 3. Click Purge. 4. In the Metric Purge panel, select one of the following: Purge all periods Select this option to purge all of the periods in the metric. Select previous periods to purge Select this option to select a specific period, and then specify the number of periods to purge in the number box. Periods are defined by the calendar you selected when you created the metric. 5. Click OK.
Publishing model-based metrics in Predictive Analysis

1. In Predictive Analysis > Metrics, select a metric group from the dropdown list. 2. Select from the list the metric you want to publish. 3. Click Publish.
31
Predictive Analysis model-based metrics Publishing model-based metrics in Predictive Analysis
The Goal & metric publishing wizard appears.
Goal & Metric Publishing Wizard

Once you create a goal, use the "Goal & Metric Publishing Wizard" to distribute the goal to individuals or groups. Goals are published for the following reasons: so that the people are focused on the right goals to enable collaboration via discussion threads and previous actions to recommend actions that link to goal performance, other reports and actions
The "Goal & Metric Publishing Wizard" has the following components to help you publish a goal or metric: Strategy Who What When Actions Summary
The Strategy step of the Goal & Metric Publishing Wizard

Select strategy in which you want to create the role.
Continue to Who.
Related Topics
The Who step of the Goal & Metric Publishing Wizard on page 32
The Who step of the Goal & Metric Publishing Wizard

1. Select a reporting role for yourself. For example, you are Vice President of both Engineering and Testing, and want to send information only to Testing. You select the reporting role for Testing. Do one of the following:
32
Select a reporting group from the list. Click Select another role. In the "Reporting Role" panel, select the group in which the role exists, choose a reporting role from the list, and click OK.
2. Select the people who can see the information: If you select People in my team, then select from the following options: If you selectEverybody in my team, members in the team are displayed, along with the link Who are they that lists the members and their titles. If you select All my direct reports, all of your direct reports are displayed, along with the link Who are they that lists these people and their titles. If you select Specific people in the team, a sub-panel appears in which you can select recipients by team roles or individuals and click Add to add them to the "Selected members" panel. The link Who are they lists the selected individuals, those within the team roles, and their titles. My manager If you select Other people, then select from the following: Org. Chart Roles - list of roles that you have already created. Individuals - list of all individuals Other Roles - list of other roles within your organization.
Note: When you select a list, a list box allows you to select to see the
list by entire company or within a specific group. 3. Click Next.

Related Topics
The What panel of the Goal & Metric Publishing Wizard on page 33
The What panel of the Goal & Metric Publishing Wizard

The "What" step enables you to specify the goal slices you want to publish. Slices of a dimension can be secured.
Note: If you publish a goal based on a secured sliced metric, by default the
publication follows respects user security rules and contains the metric permissions.
33
1. Select the object that you want to send: If you select Goal, select the goal you want to publish from the scroll list.
Note: The scroll list can be populated with all of the goal slices or by
specific goal groups in the list. If a goal seems to be missing, click Refresh. You can view the analytic for the goal by clicking View Goal. If you select Metric, select the metric you want to publish from the scroll list.
Note: The scroll list can be populated with all of the metrics or by
specific metric groups listed in the list. If a metric seems to be missing, click Refresh. You can view the analytic for the metric by clicking "View Metric". 2. Click Next.
Related Topics
The When panel of the Goal & Metric Publishing Wizard on page 34
The When panel of the Goal & Metric Publishing Wizard

"Scheduled Distribution Information" indicates the schedule on which the metric or goal's metric is evaluated. Click Next.
Related Topics
The Actions step of the Goal & metric publishing wizard on page 34
The Actions step of the Goal & metric publishing wizard

1. Select the actions that you want to recommend to the recipients by clicking one and then Ctrl+click to select others. 2. Clicking Add. 3. Select an action and click Edit Action and edit the parameters of the action. Your action selection determines the default Recommended Action text.
34
If you selected... Call Vendor Check Inventory
The default text is... Call Vendor Click <a href="inventory.html" target="In ventory">Here</a> to check product inventory
Create a <a href="../sb/flowwiz Create a new publicaframe.jsp?isclose=Y&contentType=3" tar tion get="Strategy Builder">new publication</a> Type next step text here. Custom Action Note: You can use the action to activate an action based upon the status of a sliced or non-sliced goal. Investigate <a href="defineurl.html" tar get="new"> document</a> Investigate related document Note: You can pass a slice to a Desktop Intelligence document using the action. To do this, edit the URL to specify the prompt and a slice variable, either $SLICE_NAME$ or $SLICE_CODE$. For example: &nbPrompts=1&Region=$SLICE_NAME$ Investigate <a href="definetyly.html" tar get="new"> TY-LY Analytic</a>
<a href="Outlook:Calendar"> Schedule Meeting</a> with concerned parties
Investigate related TY-LY Analysis
Schedule Meeting with Concerned Par- Note: If the default text is used, Schedule Meeting ties appears as a link that opens the recipient's Outlook calendar when clicked.
<a href="mailto:?subject= $strate gy.Name$">Send Email</a> to concerned parties
Send Email
Note: You can activate an action based on the status of a sliced or
non-sliced goal.
35
4. To personalize the Recommended Action text, change the text manually or click Personalize. In the Personalize panel, select attributes by Strategy, Role, and User, and click OK. 5. Select a Help Insert Link if you want to insert a link: To a document, for example a vendor call sheet for the Vendor Call action. Select a link from the All Documents list or from a Sub folders document list. To a web page, for example to a sales vendor site. Type the: URL To personalize the text, click Personalize. In the Personalize panel, select attributes by strategy, role, and user, and click OK. Link text Target
6. Click Insert Link, then OK. 7. The actions are executed in sequential order. To change the order of the actions, use Move Up and Move Down. 8. On each action, set the recommendation action: If the action must go with any status, select Always. If the action must go only when the status is a specific color, select Only when status is and the color of the status. 9. Click Next.
Related Topics
The Summary panel of the Goal & metric publishing wizard on page 36
The Summary panel of the Goal & metric publishing wizard

The "Summary" panel shows the settings you have chosen for the goal or metric that you are publishing. 1. To change the strategy, click the selection and the wizard returns to the "Strategy" panel. 2. The name of the report defaults to the name of the metric or goal that you selected. To change this name, edit the text in the "Name" box. 3. Optionally, type a description to go with the goal or metric.
36
4. Select the priority. 5. To modify the recipients, click the selection and the wizard returns to the "Who" panel. 6. To change the document shown, click the selection and the wizard returns to the "What" panel. 7. To modify the advanced options, click Advanced. In the "Advanced Options" panel, you have the following options: Select Send email notification to inform the goal recipients when the metric is refreshed and that there is a "Goal Activity" in their "My Goals" inbox. To modify the amount of flow and goal activities kept in the recipient's "My Goals" inbox, change the setting in the "Quantity kept" box. This setting is for the "Flow Subscription" documents. For example, if the quantity kept is two, the last two goal activities, triggered by the last two metric refreshes, are always kept for the recipient. Subsequent refreshes delete the oldest entry of the two previous goal activities. Activate Share actions for all recipients if you want to share actions. Otherwise the default is Maintain as individual. All goal subscriptions can have some actions associated with them. When actions are shared between the recipients then they can view the actions taken by the other recipients. Shared actions are good for team-oriented tasks, and the recipients can divide the actions among themselves. Maintain as individual implies that a recipient cannot view the actions taken by the other recipients. Individual actions are tasks that a goal recipient completes independently. Click OK. 8. Click Create Publication.
Related Topics
The Strategy step of the Goal & Metric Publishing Wizard on page 32 The Who step of the Goal & Metric Publishing Wizard on page 32 The What panel of the Goal & Metric Publishing Wizard on page 33
37
Predictive Analysis model-based metrics The model-based metrics list in Predictive Analysis
The model-based metrics list in Predictive Analysis

You can view the model-based metrics list in the following ways: Flat list Select this option to see a flat list of the metrics. By Measures Select this option to see the list of metrics by the measures on which they are based. By Sets Select this option to see the list of populations by sets.
Note: This option is only available if the metric is based on a set universe.
Show Groups Select this option to see the list by set groups.
By subsets Select this option to see the list by subsets, for example by joiner or leaver subsets.
38
Predictive analytics
Predictive analytics Selecting a predictive model to analyze in a Predictive analytic
Selecting a predictive model to analyze in a Predictive analytic

1. In Dashboard Builder > Create New Analytic, select an analytic the "Predictive Analytics" category. 2. Select the universe on which the model exists. 3. Select the model.
Related Topics
Model Gains Chart analytic based on predictive models on page 102 Key Influencers analytic based on predictive models on page 48 Influencer Gains Chart analytic based on predictive models on page 46 Influencer Detail analytic based on predictive models on page 44 Goal-based Influencer Detail analytic based on predictive models on page 41
Predictive analytic samples in the Analytics Catalog

The analytics included in Predictive Analysis samples in the Analytic Catalog are based on the following analytics: Influencer Analytics Goal-Based Influencer Detail Influencer Detail Influencer Gains Chart Key Influencers Model Gains Chart Variable Profile Box Plot Lists and Forecasts Metric Forecaster Individual List
For information on working with these analytics, see the Predictive Analysis documentation.
40
Predictive analytics Goal-based Influencer Detail analytic based on predictive models
Related Topics
Goal-based Influencer Detail analytic based on predictive models on page 41 Individual List based on predictive models on page 43 Influencer Detail analytic based on predictive models on page 44 Influencer Gains Chart analytic based on predictive models on page 46 Key Influencers analytic based on predictive models on page 48 Metric Forecaster analytic based on metrics on page 50 Model Gains Chart analytic based on predictive models on page 102 Variable Profile Box Plot based on predictive models on page 105
Goal-based Influencer Detail analytic based on predictive models

The Goal-based influencer detail analytic in Predictive Analysis sheds light on the precise nature of the relationship of a variable with a key outcome or goal. The Goal-based influencer detail analytic provides: a histogram display of statistics calculated by the same modeling engine that supports the Key influencers analytic. a detailed view of the relationship between a single influencer variable and a single outcome or goal.
You use goal-based influencer detail when you want to: identify "hot spots" (areas with unusually high or low statistic values) in the relationship between a variable and an outcome measure or indicator. profile variables that show up as important in Key influencers, in terms of descriptive statistics by bins.
With goal-based influencer detail, you can answer questions like: Where are the key breaks in the Employee Tenure variable and attrition rates are highest? What age groups account for the high importance of age as an influencer of purchase frequency? What most influences equipment downtime and throughput? How?
Note: The Key influencers analytic (what) and the goal-based influencer
detail analytic (how) answer this question.
41
Predictive analytics Goal-based Influencer Detail analytic based on predictive models
Related Topics
Key Influencers analytic based on predictive models on page 48 What is an influencer variable? on page 119
Configuring the Goal-based Influencer Detail analytic

1. Select a predictive model. 2. Select the influencer and goal you want to display. 3. Select a model metric you want to display and click >>. The metric appears in the Selected Metrics list. 4. Select the display options. 5. Click OK to save your configuration.
Related Topics
Selecting a predictive model to analyze in a Predictive analytic on page 40 Setting the display options in Goal-based Influencer Detail and Influencer Detail analytics on page 46 What is a goal? on page 117 What is a metric? on page 121 What is an influencer variable? on page 119
Setting the display options in Goal-based Influencer Detail and Influencer Detail analytics
1. Select the order in which you want the list sorted. 2. Select a binning option.
Option Base binning Description Select to have the final binning generated after the model is generated.
42
Predictive analytics Individual List based on predictive models
Option
Description A Predictive model is based on a population, influencers and goals. Under derived variables you can define user-defined binning for influencers. For example, you can define a binning for Age by dividing Age into five ranges [0-25], [25-40], [40-60],[60-80],[80100]. This information is used during generation of a predictive model. If no user-defined binning is provided then one is generated automatically. Note: If you are configuring the Influencer Detail analytic you need to select the goal for which you want final binning.
Final binning for goal
Related Topics
Goal-based Influencer Detail analytic based on predictive models on page 41 Influencer Detail analytic based on predictive models on page 44 What is binning? on page 116
Individual List based on predictive models

The Individual list analytic assembles a set of information for a group of individuals that includes model scores and derived variables. Use individual list when you want to apply a predictive model to the members of a set, or to individuals who fulfill specific criteria for targeting, special handling, and/or integration with external systems. As with other analytics, an Individual List refresh can be scheduled to occur at regular intervals. The list can be displayed and stored in XML format. With individual list, you can answer questions like: Which employees are the best candidates for senior management?
43
Predictive analytics Influencer Detail analytic based on predictive models
Related Topics
What is a derived variable? on page 116 What is a predictive model? on page 122
Configuring the Individual List analytic

1. Select a population and filter, if necessary, then click Next. 2. To select an item from the "Available objects" list, click it and then >>. You can add extra columns, and select to display only a specific amount of rows. 3. To add a column from another model, click Add extra column, select an object and click OK. The object appears in the "Selected objects" list.
Note: To remove an object from the "Selected objects" list, select the
object and click <<. 4. To restrict the number of rows in the Individual List, activate Only first and enter the number of rows in the "Rows" text box or navigate to the number using the + or -. 5. Click OK.
Influencer Detail analytic based on predictive models

The Influencer detail analytic is identical to the Goal-based influencer analytic detail in that the categories ("bins") can be based on statistical optimization with respect to a key outcome or "goal." However, it is different from the Goal-based influencer detail analytic in that it profiles a variable by subrange (or bin) based on measures defined in a Business Objects universe. The Influencer detail analytic exposes goal-based, bin-level statistics. The following deviation measures influence the analytic: goal deviation signed goal deviation weighted goal deviation
The following goal aggregates influence:
44
Predictive analytics Influencer Detail analytic based on predictive models
goal rate mean goal value goal standard deviation
You can profile an influencer based on: measures in the Business Objects universe (Influencer Profile analytic) performance metrics with respect to one or more outcomes (goals) (Influencer detail analytic) user-defined binning or model-derived binning, optimized for a selected goal. Profile performance on Goal A based on subgroups optimized for Goal B.
Note: Influencer analytics can display optimal binnings to reveal key
breakpoints in influencer-goal relationships.

Related Topics
What is a goal? on page 117 What is binning? on page 116 What is an influencer variable? on page 119
Configuring the Influencer Detail analytic

1. Select a predictive model. 2. Select the influencer and goal you want to display. 3. Select a model and subject metric you want to display and click >>. The metric appears in the "Selected Metrics" list. 4. Select the display options.
Related Topics
Selecting a predictive model to analyze in a Predictive analytic on page 40 Setting the display options in Goal-based Influencer Detail and Influencer Detail analytics on page 46 What is a goal? on page 117 What is a metric? on page 121 What is an influencer variable? on page 119
45
Predictive analytics Influencer Gains Chart analytic based on predictive models
Setting the display options in Goal-based Influencer Detail and Influencer Detail analytics
1. Select the order in which you want the list sorted. 2. Select a binning option.
Option Base binning Description Select to have the final binning generated after the model is generated. A Predictive model is based on a population, influencers and goals. Under derived variables you can define user-defined binning for influencers. For example, you can define a binning for Age by dividing Age into five ranges [0-25], [25-40], [40-60],[60-80],[80100]. This information is used during generation of a predictive model. If no user-defined binning is provided then one is generated automatically. Note: If you are configuring the Influencer Detail analytic you need to select the goal for which you want final binning.
Final binning for goal
Related Topics
Goal-based Influencer Detail analytic based on predictive models on page 41 Influencer Detail analytic based on predictive models on page 44 What is binning? on page 116
Influencer Gains Chart analytic based on predictive models

The Influencer gains chart analytic depicts the relationship between an influencer variable and a binary outcome or goal using a standard cumulative lift graph.
46
Predictive analytics Influencer Gains Chart analytic based on predictive models
As with Influencer detail and Goal-based influencer detail, the influencer gains chart analytic profiles a variable by subrange, or bin, but it does so in a manner that takes into account both the outcome variable and bin frequency or count. Influencer bins are placed in descending order of response rate along the horizontal axis. The model line shows the cumulative percent of signals achievable by targeting an increasing percent of the total population. Signal or response is assumed to be the least frequent value. With the influencer gains chart, you can answer questions like: Is the high importance of age as an influencer of monthly email campaign response based (a) on spikes in one or two specific age bands, (b) on the differential between two large age bands, or (c) on a gradual increase or decrease in the response rate with increasing age?
You use the influencer gains chart when you want to: identify "hot spots" (areas with unusually high or low statistic values) in the relationship between a variable and an outcome measure or indicator in a manner that encompasses "percent of population accounted for." profile variables that show up as important in Key Influencers, in terms of Cumulative Lift, the basis for "importance" calculation. identify the best subranges of a key influencer variable to use in segment creation.
Configuring the Influencer Gains Chart analytic

1. Select a predictive model. 2. Select the influencer and goal. 3. Select a model metric and click >>. The metric appears in the "Selected Metrics" list. 4. Select metric options:
Option Description
Final binning for Select this option to have the final binning generated after a goal the model is generated.
47
Predictive analytics Key Influencers analytic based on predictive models
Option Base binning
Description A Predictive model is based on a population, influencers and goals. Under derived variables you can define userdefined binning for influencers. For example, you can define a binning for Age by dividing Age into 5 ranges [0-25], [25-40], [40-60],[60-80],[80-100]. This information is used during generation of model. If no user-defined binning is provided then one will automatically generated. Select this option to display the best maximum lift response first for each bin. Select this option to organize bins based on the model results.
Best gain Model gain
Related Topics
Selecting a predictive model to analyze in a Predictive analytic on page 40 What is an influencer variable? on page 119 What is a predictive model? on page 122 What is binning? on page 116
Key Influencers analytic based on predictive models

Key influencers graphically displays the importance of different variables as an influencer, or predictor of one or more key outcomes. Users can configure key influencers to display either a simple "variable importance" metric or a "net relevance" metric that nets out the influence of correlated variables. You use key influencers when you want to: show which measures or dimensions best predict key outcomes. determine if the same or different variables are driving different types of outcomes.
The analytic exposes goal-based, influencer-level statistics. The following measures influence the analytic: Importance
48
Predictive analytics Key Influencers analytic based on predictive models
Importance represents a measurement of how well the variable, taken alone, anticipates values of the goal. It is displayed in % of maximum possible value. Obsolescence Obsolescence is an advanced user metric. It is calculated based on how much importance has changed since the last time the model was generated. Values over five suggest that the model needs to be regenerated to reoptimize the variable goal-based binning. Net relevance Net relevance is an advanced metric that represents the utility of the variable for anticipating goal values, net of other influencers in the model with over-lapping effects. It is the best metric to use for prioritization when simplifying a model. However, care must be taken, since a variable with high importance can show up as having low net relevance if it is highly correlated with another variable whose power was slightly higher in the most recent refresh (random variation in a subsequent refresh can cause two such variables to trade places). With key influencers, you can answer questions like: Which attributes and behavior are most associated with high purchase volume? Which are most associated with high risk of attrition? Where are the influencers of risk and revenue outcomes the same and where are they different? Which dimensions or measures should I use for high-value, high-potential, or high-risk segment creation? What factors determine the success of a sales professional? What most influences equipment downtime and throughput? How?
Note: Key influencers (what) and the Goal-based influencer detail analytic
(how) answer this question.

Related Topics
Goal-based Influencer Detail analytic based on predictive models on page 41
49
Predictive analytics Metric Forecaster analytic based on metrics
Configuring the Key Influencers analytic

1. In the analytic edit panel, expand Select Model to analyze and select the universe on which the model exists. 2. Select the predictive model from the "Selected Model" list. If the model does not appear, click the refresh icon next to "Selected Model". The influencers and goals attributed to the model appear in the "Associated influencers" and "Selected Goal(s)" lists. 3. Select an influencer level from the "Metric to display" list. 4. In the "Display Options" section, configure the following options: If you do not want influencers that exceed the obsolescence of your choice visible on the list, activate Hide influencers if obsolescence more than and type a limit. If you want the influencers listed in a particular order, activate Sort in X order of X for goal X and choose parameters. To create a hyperlink from the legend to a document, activate Use Legend as hyperlink to go to and browse to the document. 5. Click OK to save your changes.
Related Topics
Key Influencers analytic based on predictive models on page 48
Metric Forecaster analytic based on metrics

The Metric forecaster analytic: graphically displays metric history and forecasts one or more points into the future. automatically detects and adjusts for cyclicality, such as recurrent day-of-week or day-of-month patterns. uses error bands and retrospective 1-point-forecasts help the user judge forecast reliability.
You use metric forecaster when you want to: without needing any statistical knowledge, create forecast time-series metrics multiple points into the future that are automatically updated.
50
gain insight into forecast accuracy via error bands and retrospective forecast overlay.
With the metric forecaster analytic, you can answer questions like: Is demand likely to reach levels that would exceed planned supply at any end-of-month peak over the next 90 days? Are the sales representatives in danger of missing sales targets?
Note: You can test all combinations of detrending method and cyclicality
encoding, and select the best combination based on the following criteria: Eliminate models where Corr (predicted, actual) is not statistically significant Select the model with smallest root mean squared error:
RMSE = sqrt(sum((forecasti-actuali)2)/N)
Keep in mind
You need at least 18 periods to have a correct projection. Metric slices cannot be used with this analytic.
Related Topics
What is a metric forecaster detrended mean? on page 101
Configuring the Metric Forecaster analytic

You can use a model-based metric for this analytic. Model-based metrics are created in Predictive Analysis. For more information see the Predictive Analysis documentation. 1. In the analytic edit panel, expand Forecast parameter and click Select Metric. 2. Select a model-based metric. 3. To be able to select a different model-based metric each time the analytic is refreshed, activate Add a Prompt and type prompt text you want to appear in the prompt window. The prompt always defaults to the last model-based metric selected using the prompt. 4. Specify the number of forecast periods. 5. Expand Display Options and type a title for the analytic.
51
6. To set a hyperlink to a Central Management Server analytic from the chart legend, activate Use Legend as hyperlink to go to and click Browse to locate the analytic. The hyperlink text appears in the text box in an openAnalytic text string. You can edit this text string. For information on openAnalytic, see the Linking from goal, universe query and metric-based analytics section in the documentation. 7. Select an error band width: Select Liberal to have the error band computed as +1.96 times L2 to -1.96 times L2. Select Conservative to have the error band computed as +3 L2 to -3 L2 where L2 is root mean squared error. The error band displays the margin of error allowed. 8. Select a sampling data set to divide a population into specific parts to use in the model generation process. Select Complete to use three quarters of the data for estimation and one quarter of the data for validation. Select Optimized to use all of the data for estimation and the last two thirds of the data for validation. The default mode, "Complete", works better for smaller data sets and metrics with trends. "Optimized" mode works better for metric with cycles.
Linking from goal, universe query and metric-based analytics

Linking to multiple documents from an analytic based on a goal, metric or universe query
You can create links to multiple documents from a goal or metric-based analytic. This is particularly useful for analytic drill down. 1. How you access the options for linking analytics to other documents, depends on whether you are creating a new analytic or editing an existing analytic. If you are creating an analytic, do one of the following in InfoView: Go to Open > Dashboard Builder > Create New Analytic. Go to Document List > New > Analytic. If you are editing an analytic, select the analytic, then click Edit.
2. Expand Navigation.
52
3. The next step depends on whether you want: If you want to link the title of the analytic to multiple documents, click Multiple Links below "Use Legend/Title as hyperlink to go to". If you want to provide data point navigation, so that users can click a metric result and then view a filtered document that shows other results for that specific time period, click Multiple Links below "Use data point as hyperlink to go to". The Edit Navigation Link dialog box appears. 4. Click Add, then click Browse to select the document or, if you want to link to a web page, type the URL into the text box. 5. You can add several links to documents here. 6. Click Update, then click OK. The link to the multiple documents appears. For example:
Drill Down Group Level||openAnalytic.jsp?Document Name=Drill+Down+Group+Level&RepositoryType=C&Repository Name=Feature Examples&DocumentExt=rpt&DocumentId=AWbJO JEYvvVMnbtXEQ2An0s&sContentType=OnDemand&mode=full||Total Sales.rpt||openAnalytic.jsp?DocumentName=TotalSales.rpt&Repos itoryType=C&RepositoryName=Feature Samples&Docu mentExt=rpt&DocumentId=AdLEddSQNTlDhMiLiHoD7II&sContent Type=OnDemand&mode=full||Business Objects||http://www.busi nessobjects.com
In this example the multiple link is created to: Drill Down Group Level TotalSales report www.businessobjects.com The syntax of the links is as follows:
name1||URL||name2||URL||name3||URL
7. To check the links, click OK. The analytic appears. 8. Place your cursor on a data point or the title. If the links are defined correctly, a popup menu appears with list of destination documents.
53
9. Click the destination document of your choice.

Note: When linking from goal and predictive analytics you must enter
the link manually.

Related Topics
OpenAnalytic variables used in goal, metric and universe query analytics on page 55 Using variables for dynamic links to documents on page 55
Linking from a goal, metric or universe query-based analytic to a document or analytic

Adding links to analytics allows end users navigate from an analytic to other analytics or documents that provide information for further analysis. You can include prompt parameters in links. Adding prompt parameters means that the data displayed on the target analytic or document is dynamically generated according to the values displayed on the first analytic. 1. How you access the options for linking analytics to other documents depends on whether you are creating an analytic or editing an existing analytic: If you are creating an analytic, do one of the following in InfoView: Go to Open > Dashboard Builder > Create New Analytic. Go to New > Analytic. If you are editing an analytic, select the analytic, then click Edit.
2. Expand Navigation. 3. The next step depends on whether you want to link the title or data points to a destination document: If you want to link the title of the analytic to multiple documents, click Browse below "Use Legend/Title as hyperlink to go to". If you want to provide data point navigation, so that users can click a metric result to view a filtered document that shows other results for that specific time period, click the Browse below "Use data point as hyperlink to go to". The "Edit Navigation Link Menu" dialog box appears. 4. Click Add, then Browse to select the document or, if you want to link to a web page, type the URL into the text box.
54
Note: When linking from goal and predictive analytics you must enter
the link manually. 5. Click Update, then click OK. The URL to the document appears in the URL box. The URL reads as follows:
openAnalytic.jsp?DocumentName=<FILE_NAME> &RepositoryType= C&RepositoryName=<REPOSITORY_NAME> &DocumentExt=<EXTN>& DocumentId=<DOCID> &sContentType=OnDemand&mode=full
6. To check the links, click OK. The analytic appears. 7. Place your cursor on a data point or the title. 8. When your cursor turns into a hand sign, click the data point or title. The destination document appears.
Related Topics
OpenAnalytic variables used in goal, metric and universe query analytics on page 55 Linking to multiple documents from an analytic based on a goal, metric or universe query on page 52 Using variables for dynamic links to documents on page 55 Using variables for dynamic links to documents on page 55
Using variables for dynamic links to documents

You can include variables in the links to documents so that the destination document is filtered by the values for a specific metric, dimension, goal or set.
Related Topics
OpenAnalytic variables used in goal, metric and universe query analytics on page 55
OpenAnalytic variables used in goal, metric and universe query analytics

The following table gives a description of each variable you can use when used in creating links between analytics.
55
Note: A universe query analytic where a prompt is defined can use all of the
OpenAnalytic variables.
Variable $METRIC_ID$ $METRIC_NAME$ $SUBJECT_ID$ $SUBJECT_NAME$ $SET_ID$ $SET_NAME$ $SET2_ID$ $SET2_NAME$ $POPULATION_ID$ $POPULATION_NAME$ $DIM_ID$ $DIM_NAME$ $DIM2_ID$ $DIM2_NAME$ $GOAL_NAME$ $GOAL_ID$ $SLICE_CODE$ $SLICE_NAME$ $CONDITION_ID$ Description Metric ID Metric Name Subject ID Subject name Set ID Set Name 2nd Set ID (when metric based on two sets) 2nd Set Name (when metric based on two sets) Population ID Population Name Dimension ID Dimension Name 2nd Dimension ID 2nd Dimension Name Name of goal associated with analytics ID of goal associated with analytics Value of the slice (code) Value of the slice (name) Class Id from the source universe
56
Variable $MEASURE_NAME$ $MIN_SLICE$ $MAX_SLICE$ $CURRENT_SLICE$ $BEGIN_DATE$ $END_DATE$ $CURRENT_DATE$ $MAP_CODE$ $MAP_NAME$ $MAP_LONGNAME$ $MODEL_ID$ $INFLUENCER_ID$ $INFLUENCER_NAME$
Description Measure Name First slice of the dimension used for x-axis (QOU) Last slice of the dimension used for x-axis (QOU) Current slice of the dimension used for x-axis (QOU) Begin Date of the displayed period End Date of the displayed period Current period (corresponds to a data point in a trend) Code for the current region displayed Name of current region displayed Long name of current region displayed Model ID Influencer ID Influencer Name
OpenAnalytic variables for Goal variance and Pareto analytics

Pareto analytic
You can only use the $METRIC_ID$ variable to link from a Pareto analytic.
Goal variance analytics
You can use the following variables to link from Goal Variance analytics: $CONDITION_ ID$ $CONDITION_ NAME$
57
$DIM_ID$ $DIM_ NAME$ $GOAL_ ID$ $MEASURE_ NAME$ $METRIC_ ID$ $METRIC_ NAME$ $POPULATION_ ID$ $POPULATION_ NAME$ $SET_ ID$ $SET_ NAME$ $SET2_ ID$ $SET2_ NAME$ $SLICE_ CODE$ $SLICE_ NAME$ $SUBJECT_ ID$ $SUBJECT_ NAME$ $USER$
Related Topics
OpenAnalytic variables for Map analytics

This table shows which variables can be used to link from Map analytics that use metrics or a universe query.
58
Map analytic using metrics Variable Title link $METRIC_ ID$ X Map item link X X X X X X X X X
Map analytic using a universe query Title link X X X X X X X X X Map item link X X X X X X X X X
$METRIC_ NAME$ X $SUBJECT_ ID$ $SUBJECT_ NAME$ $SET_ID$ $SET_ NAME$ $SET2_ID$ $SET2_ NAME$ $POPULATION_ ID$ $POPULATION_ NAME$ $DIM_ID$ $DIM_NAME$ $GOAL_NAME$ $GOAL_ID$ $SLICE_CODE$ $SLICE_NAME$ X X X X X X X X X
X X X
X X X
X X X
X X X X
X X X X X
X X X X
$CONDITION_ ID$ X
59
Map analytic using metrics Variable Title link $CONDITION_ NAME$ $MEASURE_ NAME$ $BEGIN_DATE$ $END_DATE$ $MAP_CODE$ $MAP_NAME$ $MAP_ LONGNAME$ $USER$ X X X X X X Map item link X
Map analytic using a universe query Title link Map item link
X X X X X
X X X X X X
Related Topics
OpenAnalytic variables for Metric List and Strategy Map analytics

This table shows which variables can be used to link from Metric Lists and Strategy Maps analytics (only when based on goals, metrics or universe queries).
Metric List Variable Title link $METRIC_NAME$ X Attached analytic X Strategy Map
60
Metric List Variable Title link $SUBJECT_ID$ $SUBJECT_ NAME$ $SET_ID$ $SET_NAME$ $SET2_ID$ $SET2_NAME$ $POPULATION_ ID$ $POPULATION_ NAME$ $DIM_ID$ $DIM_NAME$ $GOAL_NAME$ $SLICE_CODE$ $SLICE_NAME$ $CONDITION_ID$ $CONDITION_ NAME$ $MEASURE_ NAME$ $USER$ X X X X X X X X X X X X X X X X X
Strategy Map Attached analytic X X X X X X X X X X X X X X X X X
61
Related Topics
OpenAnalytic variables for Metric Tree and Metrics Overview analytics

This table shows which variables can be used to link from Metric Trees and Metrics Overview analytics.
Metric Tree Variable Metrics Overview Datapoint Link
Attached Analyt- Set Name ic Link X X X X X X X X X X X X X X X X
$METRIC_ID$ $METRIC_NAME$ $SUBJECT_ ID$ $SUBJECT_ NAME$ $SET_ID$ $SET_NAME$ $POPULATION _ID$ $POPULATION _NAME$ $DIM_ID$ $DIM_NAME$ $GOAL_NAME$ $GOAL_ID$ $SLICE_ CODE$
X X X X X X X X X
62
Metric Tree Variable
Metrics Overview Datapoint Link X X X X X X X
Attached Analyt- Set Name ic Link X X X X
$SLICE_ NAME$ $CONDITION _ID$ $CONDITION _NAME$ $MEASURE _NAME$ $BEGIN_DATE$ $END_DATE$ $CURRENT _DATE$ $USER$
Related Topics
OpenAnalytic variables for most volatile metrics and Set Membership analytics
This table shows which variables can be used to link from most volatile metrics and set membership analytics.
Most Volatile Metrics Variable Analytic Link $METRIC_ID$ $METRIC_NAME$ $SUBJECT_ID$ X X X Datapoint X X X X Datapoint Set Membership
63
Most Volatile Metrics Variable Analytic Link $SUBJECT_NAME$ $SET_ID$ $SET_NAME$ $SET2_ID$ $SET2_NAME$ $POPULATION_ID$ $POPULATION_NAME$ $DIM_ID$ $DIM_NAME$ $SLICE_CODE$ $SLICE_NAME$ $CONDITION_ID$ $CONDITION_NAME$ $MEASURE_NAME$ $BEGIN_DATE$ $END_DATE$ $CURRENT_DATE$ $USER$ X X X X X X X X X X X X X X X X X X Datapoint X X X X X X X X X X X X X X X X X X
Set Membership Datapoint X X X
64
Related Topics
OpenAnalytic variables for Key Influencers analytics

The following variables can be used to link from Key Influencers analytics: $MODEL_ID$ $INFLUENCER_ID$ $INFLUENCER_NAME$
Related Topics
OpenAnalytic variables for the Metric Forecaster analytic

The following variables can be used to link to Metric Forecaster analytic: $METRIC_ ID$ $BEGIN_ DATE$ $END_ DATE$
OpenAnalytic variables for the Interactive Metric Trend analytic

This table shows the variables can be used to link from Interactive Metric Trends based on goals, metrics or universe queries.
Variable $METRIC_ID$ $METRIC_NAME$ $SUBJECT_ ID$ $SUBJECT_ NAME$ $SET_ID$ Legend X X X X X Datapoint X X X X X
65
Variable $SET_NAME$ $SET2_ID$ $SET2_NAME$ $POPULATION_ ID$ $POPULATION_ NAME$ $DIM_ID$ $DIM_ NAME$ $SLICE_ CODE$ $SLICE_ NAME$ $CONDITION_ ID$ $CONDITION_ NAME$ $MEASURE_ NAME$ $MIN_ SLICE$ $MAX_ SLICE$ $CURRENT_ SLICE$ $USER$ $BEGIN_DATE$ $CURRENT_DATE$ $END_DATE$
Legend X X X X X X X X X X X X X X
Datapoint X X X X X X X X X X X X X X X
X X
X X X
66
Related Topics
OpenAnalytic variables for a Migration analytic

The following variables can be used to link from a datapoint in the Migration analytic for set analysis: $SUBJECT_ID$ $SUBJECT_ NAME$ $SET_ID$ $SET_NAME$ $USER$
Related Topics
OpenAnalytic variables for Gauge analytics

This table shows which variables can be used to link from Gauges.
Variable $METRIC_ ID$ $METRIC_ NAME$ $SUBJECT_ ID$ $SUBJECT_ NAME$ $SET_ID$ $SET_NAME$ $SET2_ID$ $SET2_NAME$ Metric X X X X X X X X Universe query
67
Variable $POPULATION _ID$ $POPULATION _NAME$ $DIM_ID$ $DIM_NAME$ $DIM2_ID$ $DIM2_NAME$ $GOAL_NAME$ $GOAL_ID$ $SLICE_ CODE$ $SLICE_ NAME$ $CONDITION _ID$ $CONDITION _NAME$ $MEASURE _NAME$ $MIN_SLICE$ $MAX_SLICE$ $CURRENT _SLICE$ $BEGIN_DATE$ $END_DATE$ $CURRENT _DATE$ $USER$
Metric X X X X
Universe query
X X X X
X X X X X X X X X X X X X X X X
68
OpenAnalytic variables for the Pie, Bubble and Radar Charts

The following table shows which variables can be used to link from the Pie, Bubble and Radar charts:
Variable $METRIC_ ID$ Pie Chart X Bubble Chart X X Radar Chart
$METRIC_ NAME$ X $GOAL_ ID$ $GOAL_ NAME$ $BEGIN_ DATE$ $END_ DATE$ $CURRENT_ DATE$ X X X X X
X X X
X X X
Related Topics
OpenAnalytic variables for Data Exploration analytics

The following variables can be used to link from the Data Exploration analytics, which include Lag Plot, Frequency Histogram, Normal Probability Plot, Smoothing and Curve Fitting : $METRIC_ID$ $METRIC_NAME$ $BEGIN_DATE$ $END_DATE$ $CURRENT_DATE$
69
Related Topics
nbPrompts
nbPrompts is a parameter used in customized links between analytics and documents stored on the Central Management Server in which prompts are allowed, including analytics, Crystal Reports, Desktop Intelligence and Web Intelligence documents. Using nbPrompts, you can fill the prompt(s) in the target document with openAnalytic variables.
nbPrompts syntax in links from goal and metric-based analytics

The following are examples of nbPrompts syntax:
openanalytic.jsp?...&mode=full&nbPrompts=1 &[analytic_prompt_name]=$METRIC_NAME$ openanalytic.jsp?...&mode=full&nbPrompts=3 &[analytic_prompt1]=$METRIC_NAME$&[analytic_prompt2]= $BEGIN_DATE$&[analytic_prompt3]=$END_DATE$
where openanalytic.jsp?...&mode=full is automatically included in the address when you select the target document, and you add the remaining text using the following syntax:
&nbPrompts=[#]&[analytic_prompt_name]= $METRIC_NAME$
where you replace [#] with the number of prompts, and for each prompt you add the string &[analytic_prompt_name]=$METRIC_NAME$, where [analytic_prompt_name] is replaced with the name of the prompt in the analytic or document.
Using openDocument in dashboards and analytics

The openDocument.jsp uses the "opendoc" context to open objects from a document or dashboard in InfoView. Use the following syntax:
/OpenDocument/opendoc/ openDocument.jsp?iDocID=vscmMCXk &sIDType=CUID&sType=null&sInstance=Last
70
where vscmMCXk is the ID of the dashboard.

Note: Use relative instead of absolute links to documents in InfoView, in the
case where you plan to migrate servers. To use openDocument to open a dashboard from a document or dashboard:
/OpenDocument/opendoc/openDocument.jsp? iDocID=vscmMCXk&sIDType=CUID&sType=null
To use openDocument to open an analytic from a document or dashboard:

/OpenDocument/opendoc/openDocument.jsp? iDocID=vscmMCXk&sIDType=CUID&sType=null
Example: sInstance in openDocument syntax
Use the parameter sInstance=Last with the OpenDocument function to retrieve the last instance of a Web Intelligence or Desktop Intelligence report. You must specify the fully-qualified URL of the target document for the parameter to take effect.
Note: sInstance=Last has no effect in the OpenAnalytic function.
URL Reporting Using openDocument

URL reporting using openDocument provides URL access to multiple document types by passing a URL string to a BusinessObjects Enterprise server. openDocument provides commands to control how reports are generated and displayed. You can use openDocument in BusinessObjects Enterprise to create cross-system links to and from the following document types: .wid: Web Intelligence version 6.x documents .rep: Desktop Intelligence documents .rpt: Crystal reports .car: OLAP Intelligence reports
Note: For more information about how to customize your Web Intelligence
documents, see the Web Intelligence Report Engine Developer Guide.
71
Structuring an openDocument URL

The next sections explain how to use the openDocument function, and how to construct the URL. An openDocument URL is generally structured as follows:
http://<servername>:<port>/OpenDocument/<platformSpecific>?<pa rameter1>&<parameter2>&...&<parameterN>
The exact syntax of the <platformSpecific> parameter depends on your implementation: For Java implementations, use opendoc/openDocument.jsp in place of the <platformSpecific> parameter.
The URL is constructed using the parameters listed in OpenDocument parameter overview on page 75
Joining parameters
Join parameters with the ampersand (&). Do not place spaces around the ampersand. For example: sType=wid&sDocName=Sales2003 The ampersand is always required between parameters.
Spaces and special characters in parameter values
Because some browsers cannot interpret spaces, the parameters of the link cannot contain spaces or other special characters that require URL encoding. To avoid the misinterpretation of special characters, you can define a URLEncoded string in the source database to replace the special character with an escape sequence. This will allow the database to ignore the special character and correctly interpret the parameter value. Note that certain RDBMS have functions that allow you to replace one special character with another. By creating an escape sequence for the plus sign (+), you can instruct the database to interpret the plus sign as a space. In this case, a document title Sales Report for 2003 would be specified in the DocName parameter as:
&sDocName=Sales+Report+for+2003&
This syntax prevents the database from misinterpreting the spaces in the title.
72
Trailing spaces in parameter values
Trim trailing spaces at the end of parameter values and prompt names. Do not replace them with a plus sign (+). The viewer may not know whether to interpret the plus sign (+) as part of the prompt name or as a space. For example, if the prompt name displays:
Select a City:_
(where _ represents a space), enter the following text in the link:

lsSSelect+a+City:=Paris
where the spaces within the prompt name are replaced with the plus sign, and the trailing space is trimmed off. For details on prompt parameters of the link, refer to OpenDocument parameter overview on page 75 .
Capitalization
All of the openDocument parameters are case sensitive.

Link length limit
The encoded URL cannot exceed 2083 total characters.

Parameter values in links to sub-reports
You cannot pass parameter values to a sub-report of a target Crystal report.
Using the lsS parameter with OLAP Intelligence reports

If the target document is an OLAP Intelligence report (.car) you can use the IsS parameter to specify prompts. The parameters are passed in as a URL-encoded string using the unique name of the parameter set up in the OLAP Intelligence report.
Example: Opening a report to a specific page
If 23CAA3C1-8DBB-4CF3-BA%2CB8%2CD7%2CF0%2C68%2CEF%2C9C%2C6F is the URL-encoded unique name for the page parameter in the OLAP
73
Intelligence report, you would use the following URL to open the OLAP Intelligence report to page 2:
http://<servername>:<port>/OpenDocument/<platformSpecif ic>?sType=car&sIDType=InfoObject&iDocID=440&lsS23CAA3C1-8DBB4CF3-BA%2CB8%2CD7%2CF0%2C68%2CEF%2C9C%2C6F=2
Example: Opening a cube parameter
If 8401682C-9B1D-4850-8B%2C5E%2CD9%2C1F%2C20%2CF8%2C1%2C62 is the URL-encoded unique name for the cube parameter opening the warehouse cube in the catalogue FoodMart 2000 on MSAS, you would use the following URL to open this cube parameter:
http://<servername>:<port>/OpenDocument/<platformSpecif ic>?sType=car&sIDType=InfoObject&lsS8401682C-9B1D-48508B%2C5E%2CD9%2C1F%2C20%2CF8%2C1%2C62=CATALOG%3DFood Mart%202000,CUBE%3Dwarehouse&iDocID=616
Using the lsM parameter with OLAP Intelligence reports

If the target document is an OLAP Intelligence report (.car) you can use the IsM parameter to specify prompts. The parameters are passed in as a URL-encoded string using the unique name of the parameter set up in the OLAP Intelligence report. As was the case for the lsS parameter, lsM parameters are also passed in as a URL-encoded string using the unique name of the parameter set up in the OLAP Intelligence report.
Example: Opening a report
http://<servername>:<port>/OpenDocument/<platformSpecif ic>?sType=car&sIDType=InfoObject&lsMADC216EA-D9A5-42B5AE%2C21%2C84%2CA9%2CF9%2C6E%2C31%2C7=[%5BCustomers%5D.%5BCoun try%5D.%26%5BMexico%5D],[%5BCustomers%5D.%5BCoun try%5D.%26%5BCanada%5D]&iDocID=544
This is a memberset parameter opening up a report with Customers > Country > Mexico and Customers > Country > Canada in the view.
74
OpenDocument parameter overview

This section provides a brief overview of openDocument and includes a list of available commands. Details about the available commands, their specific uses, and relevant examples are also provided. The exact syntax of the <platformSpecific> parameter depends on your implementation: For Java implementations, use opendoc/openDocument.jsp in place of the <platformSpecific> parameter.
The first parameter of the link

http://<servername>:<port>/OpenDocument/<platformSpecific>?
The first parameter, as displayed in the example above, must precede all other parameters. After this information, the parameters can appear in any order. The parameters of the function are displayed in the following table. The mandatory column indicates whether the parameter is required in the link.
Note: The document containing the openDocument link is called the parent
document, and it resides on the parent system. The document to which the link points is called the target document, and it resides on the target system.
Table 6-11: Platform Parameters
Parameter iDocID on page 78 sDocName on page 78 sIDType on page 79 sKind on page 79
Description Document identifier. Document name. Crystal object type. The file type of target Desktop Intelligence document. The name of the folder and subfolder containing the target document.
sPath on page 80
75
Parameter sType on page 80
Description The file type of target document or report. A valid logon token for the current CMS session.
token on page 81
Table 6-12: Input Parameters
Parameter
Description Specifies a contextual prompt if there is an ambiguity during SQL generation (Business Objects and Web Intelligence documents only). Specifies a range of values for a prompt. [NAME] is the text of the prompt. For Crystal targets only, indicates whether the link should open the full target report or just the report part specified in sReportPart. Specifies a value for a single prompt. [NAME] is the text of the prompt. Indicates which specific instance of the target report to open. In Crystal Reports, a report part is associated to a data context. Indicates whether a refresh should be forced when the target document or report is opened.
lsC on page 81
lsM[NAME] on page 83
lsR[NAME] on page 86
lsS[NAME] on page 89
sInstance on page 91
sPartContext on page 92
sRefresh on page 92
76
Parameter
Description For Crystal targets only, indicates whether the link should open the full target report or just the report part specified in sReportPart. Indicates which report to open if target document is multi-report. Indicates which specific part of the target report to open.
sReportMode on page 93
sReportName on page 93
sReportPart on page 94
Table 6-13: Output Parameters
Parameter NAII on page 95
Description Forces the display of the prompt selection page. Indicates the format in which the target document is opened. Indicates the selected report viewer (CR & CA only). Indicates whether the target report will open in the current browser window or whether a new window will be launched.
sOutputFormat on page 95
sViewer on page 96
sWindow on page 97
OpenDocument Platform parameters

The following tables list the openDocument platform parameters:
Note: Variables are denoted with angle brackets. You must substitute the
proper value for these variables. For example, you must use the name of your server in place of <servername> where it is contained in the code samples below, and you must use your port number in place of <port>.
77
iDocID
Parameter syntax
Description of parameter
Mandatory?
Values accepted for parameter
iDocID
One of sDocName Document identifiDocument identifior iDocID is er. er (InfoObjectID). mandatory.
Example:
http://<servername>:<port>/OpenDocument/<platformSpecif ic>?sType=wid&sDocName=SalesReport&iDocID=2010
Note: To obtain the document ID, navigate to the document in InfoView,
hover your mouse over the document name hyperlink, and look for the ID number in the browser's status bar. You can also obtain the document ID from the Central Management Console.
sDocName
Parameter syntax
Description of parameter Document name without extension
Mandatory?
sDocName
One of sDocName If multiple docuor iDocID is Document name. ments have the same name, speci- mandatory. fy the correct document with iDocID.
Example:
http://<servername>:<port>/OpenDocument/<platformSpecif ic>?sPath=[Sales+Reports]&sDocName=Sales+in+200
78
sIDType
Parameter syntax
Mandatory?
sIDType
Yes if the target is a Crystal report or OLAP Intelligence Central Managereport (sType=rpt ment Server or =car ) in an (CMS) object idenObject Package tifier type. (otherwise, use sPath and sDoc Name )
CUID GUID RUID ParentID InfoObjectID
(default)
Example:
http://<servername>:<port>/OpenDocument/<platformSpecific>?iDo cID=2010&sIDType=CUID sKind
Parameter syntax
Mandatory?
sKind
Yes if the target is The file type of tar- a Desktop Intelliget Desktop Intelli- gence document gence document. (otherwise, use sType )
FullClient
Example:
http://<servername>:<port>/OpenDocument/<platformSpecif ic>?sKind=FullClient
79
sPath
Parameter syntax
Mandatory?
Values accepted for parameter Folder and/or subfolder:

[folder],[sub folder]
sPath
The name of the Yes if sDocName is folder and subfoldspecified and is er containing the not unique. target document.
Example:
http://<servername>:<port>/OpenDocument/<platformSpecif ic>?sPath=[Sales+Reports]&sDocName=Sales+in+2005 sType
Parameter syntax
Mandatory?
sType
The file type of tar- Yes, but ignored get document or for agnostic docu- report. ments
wid rpt car
Example:
http://<servername>:<port>/OpenDocument/<platformSpecif ic>?sType=wid
80
token
Parameter syntax
Mandatory?
Values accepted for parameter The logon token for the current CMS session.
token
A valid logon token for the current No CMS session.
Example:
http://<servername>:<port>/OpenDocument/<platformSpecif ic>?sType=wid&sDocName=Sales+in+2003&token=<logonToken>
OpenDocument Input parameters

The following tables list the openDocument input parameters:
lsC
Parameter syntax
Mandatory?
lsC
Specifies a contextual prompt if there is an ambiguity during SQL generation (Business Objects and Web No Intelligence documents only). Note: Not supported by OLAP Intelligence
A prompt value that resolves the ambiguity in the SQL generation.
81
Example:
http://<servername>:<port>/OpenDocument/<platformSpecif ic>?sType=wid&sDocName=SalesReport&iDocID=2010&lsC=Sales
82
lsM[NAME]
Parameter syntax
Mandatory?
lsM[NAME]
Specifies multiple values for a prompt. [NAME] is No the text of the prompt.
83
Parameter syntax
Mandatory?
Multiple prompt values, separated by a comma. If the target is a Crystal report, each value must be enclosed in square brackets. If the target is a OLAP Intelligence report, use the MDX WITH clause (refer to Using the lsS parameter with OLAP Intelligence reports on page 73 and Using the lsM parameter with OLAP Intelligence reports on page 74). no_value (only for optional parameters)
Note: You can remove an optional parameter from the
84
Parameter syntax
Mandatory?
Values accepted for parameter prompt by setting it to no_value in the OpenDocument query string. If you leave an optional parameter out of the OpenDocument query string, a default parameter value will be applied.
Example:
http://<servername>:<port>/OpenDocument/<platformSpecif ic>?sType=rpt&sDocName=SalesReport&lsMSe lect+Cities=[Paris],[London]
85
lsR[NAME]
Parameter syntax
Mandatory?
lsR[NAME]
Specifies a range of values for a prompt. [NAME] is the text of the prompt. No Note: Not supported by OLAP Intelligence
86
Parameter syntax
Mandatory?
A range of values for the prompt, separated by a double period (..). If the target is a Crystal report, the range must be enclosed in square brackets and/or parentheses (use a square bracket next to a value to include it in the range, and parentheses to exclude it). no_value (only for optional parameters)
Note: You can remove an optional parameter from the prompt by setting it to no_value in the OpenDocument query string. If you leave an optional parameter
87
Parameter syntax
Mandatory?
Values accepted for parameter out of the OpenDocument query string, a default parameter value will be applied.
Example:
http://<servername>:<port>/OpenDocument/<platformSpecif ic>?sType=rpt&sDocName=SalesReport&lsRTime+Period:=[2000..2004)
88
lsS[NAME]
Parameter syntax
Mandatory?
lsS[NAME]
Specifies a value for a single prompt. [NAME] is No the text of the prompt.
89
Parameter syntax
Mandatory?
A single prompt value (refer to Using the lsS parameter with OLAP Intelligence reports on page 73 and Using the lsM parameter with OLAP Intelligence reports on page 74). no_value (only for optional parameters)
Note: You can remove an optional parameter from the prompt by setting it to no_value in the OpenDocument query string. If you leave an optional parameter out of the OpenDocument query string, a default parameter value will be ap-
90
Parameter syntax
Mandatory?
Values accepted for parameter plied.
Example:
http://<servername>:<port>/OpenDocument/<platformSpecif ic>?sType=wid&sDocName=SalesReport&iDocID=2010&lsSSe lect+a+City=Paris sInstance
Parameter syntax
Mandatory?

User (Link to
latest instance owned by current user)
sInstance Last (Link to
Indicates which No (use with specific instance sDocName and of the target report lsS[NAME]) to open.
latest instance for report)
Param (Link to
latest instance of report with matching parameter values)
Example:
http://<servername>:<port>/OpenDocument/<platformSpecif ic>?sPath=[Sales+Reports]&sDocName=Sales+in+2003&sReport Part=Part1&sInstance=User
91
sPartContext
Parameter syntax
Mandatory?
sPartContext
In Crystal Reports, Yes if a value is a report part is asspecified for sRe sociated to a data portPart context.
Data context of the report part.
Example:
http://<servername>:<port>/OpenDocument/<platformSpecif ic>?sPath=[Sales+Reports]&sDocName=Sales+in+2005&sReport Part=Part1&sPartContext=0-4-0 sRefresh
Parameter syntax
Mandatory?

Y (forces the
sRefresh
Indicates whether a refresh should be forced when No the target document or report is opened.
documents refresh)
N (note that
the refresh on open feature overrides this value)
Example:
http://<servername>:<port>/OpenDocument/<platformSpecif ic>?sType=wid&sDocName=SalesReport&iDocID=2010&sRefresh=Y
92
sReportMode
Parameter syntax
Description of parameter For Crystal targets only, indicates whether the link should open the full target report or just the report part specified in sRe portPart.
Mandatory?
No (default is Full) Only applies if a value is specified for sReportPart
Full Part
sReportMode
Example:
http://<servername>:<port>/OpenDocument/<platformSpecif ic>?sPath=[Sales+Reports]&sDocName=Sales+in+2003&sReport Part=Part1&sReportMode=Part sReportName
Parameter syntax
Mandatory?
Values accepted for parameter Report name for Web Intelligence documents, subreport for Crystal Reports, pages for OLAP Intelligence reports.
sReportName
Indicates which report to open if tar- No (default is the get document is first report) multi-report.
Example:
http://<servername>:<port>/OpenDocument/<platformSpecif ic>?sType=wid&sDocName=Sales+in+2003&sReportName=First+Re port+Tab
93
sReportPart
Parameter syntax
Mandatory?
sReportPart
Indicates which specific part of the No target report to open.
Name of the report part.
Example:
http://<servername>:<port>/OpenDocument/<platformSpecif ic>?sPath=[Sales+Reports]&sDocName=Sales+in+2003&sReport Part=Part1
OpenDocument Output parameters

The following tables list the openDocument output parameters:
94
NAII
Parameter syntax
Mandatory?
Forces the display of the prompt selection page.

NAII
Y (all prompts whose values are passed with lsS, lsM or lsR are pre-selected) N (displays only the prompts whose values passed with lsS, lsM or lsR)
Note: Not supported by OLAP Intelligence
No
Example:
http://<servername>:<port>/OpenDocument/<platformSpecif ic>?sType=wid&sDocName=SalesReport&iDocID=2010&NAII=Y sOutputFormat
Parameter syntax
Description of parameter Indicates the format in which the target document is opened.
Mandatory?

H (HTML) P (PDF) E (Excel) W (Word)
sOutputFormat
No (default is HTML if the parameter is not specified in the link)
95
Example:
http://<servername>:<port>/OpenDocument/<platformSpecif ic>?sPath=[Sales+Reports]&sDocName=Sales+in+2003&sOutputFor mat=E sViewer
Parameter syntax
Mandatory?

html actx (Crystal
sViewer
Indicates the viewer that is used to No view the document.
reports only) java (Crystal reports only)
Example:
http://<servername>:<port>/OpenDocument/<platformSpecif ic>?sPath=[Sales+Reports]&sDocName=Sales+in+2003&sViewer=html
Example: Note: In order to use parameters in the URL with the ActiveX viewer, :connect must be appended to the URL, followed by the parameters.
http://<servername>:<port>/OpenDocument/<platformSpecif ic>?sPath=[Sales+Reports]&sDocName=Sales+in+2003&sViewer=ac tx:connect&IsMCountry=[Thailand],[Norway]
96
sWindow
Parameter syntax
Mandatory?

Same (current
sWindow
Indicates whether the target report will open in the current browser No window or whether a new window will be launched.
browser window)
New (new
browser window is launched)
Example:
http://<servername>:<port>/OpenDocument/<platformSpecif ic>?sType=wid&sDocName=SalesReport&iDocID=2010&sWindow=New
Contextual report linking

The openDocument feature allows you to create contextual links between Crystal reports, OLAP Intelligence reports and Web Intelligence documents. To do this, you construct a URL using the openDocument syntax and then insert the URL into a Crystal report, OLAP Intelligence report or Web Intelligence document. Contextual report linking allows report designers to specify associations for documents residing in either a Crystal Reports environment (unmanaged) or a BusinessObjects Enterprise environment (managed). Once these associations are created, users follow the resulting navigational paths embedded in the linked documents. This feature enables you to invoke Business Objects and Web Intelligence documents from Crystal Reports and vice versa. This feature relies on functionality that allows the user to do the following: Link Web Intelligence or Business Objects documents within the document domain. Link report objects in Crystal Reports.
97
Click the appropriate link to jump to that section: To insert a link into a Crystal report on page 98 To create a link to another report or document from an OLAP Intelligence report on page 98 Creating links in Web Intelligence documents on page 99
To insert a link into a Crystal report

You can use openDocument to create hyperlinks in Crystal Reports. To create a link to another report or document, use the Hyperlink Tab of the field Format Editor. 1. Open the source report in Crystal Reports. 2. Right-click the field in which you want to insert the openDocument link and select Format from the shortcut menu. 3. In the Format Editor, select the Hyperlink tab. 4. Select A website on the Internet. 5. In the Hyperlink information area, leave the Website Address field empty and click the Format Formula Editor button. 6. Enter the openDocument link in the following format:
"http://[openDocument parameters]"+{Article_lookup.Fami ly_name}
Where [openDocument parameters] are described in OpenDocument parameter overview on page 75, and the {Article_lookup.Family_name} enables the report to pass context-dependent data.
Note: Test your link in a browser window before inserting it into a report
or document. 7. Click Save and Close to leave the Formula Workshop. 8. Click OK in the Formula Editor to save the link.
To create a link to another report or document from an OLAP Intelligence report

You can use openDocument to create hyperlinks in OLAP Intelligence reports. 1. Open the source report in the OLAP Intelligence designer. 2. On the Tools menu, select Action Manager . 3. Click New to create a new action.
98
4. Enter an action name. 5. Select the area to which the action (the link) will apply. 6. Enter the openDocument link using the parameters and syntax described in this document.
Tip: Test your link in a browser window before inserting it into a report
or document. 7. 8. 9. 10. 11. 12. 13. Click OK to save the link. Close the Action Manager dialog box. Create an Analysis Button on the source report. Right-click the Analysis Button. In the drop-down menu, select Properties and then Edit . Select Launch an action . Select the action that corresponds to the openDocument link created in steps 3 through 6. 14. Click OK .
Creating links in Web Intelligence documents

You can define objects in a universe that allow Web Intelligence and BusinessObjects users to create reports whose returned values include links to other reports and documents. When these reports are exported to the repository, users can click returned values displayed as hyperlinks to open another related document stored in the document domain of the repository.You create these links using the openDocument function in the definition of an object in Designer.
More information
For full information on creating links in Web Intelligence reports, see the Building Reports Using the WebIntelligence Java Report Panel guide. You enable report linking in a universe by creating an object (the link object) whose returned values are the same as the values used as input to a prompt in an existing report (the target report).
99
The openDocument function allows the values for the link object to be returned as hyperlinks. When the user clicks the hyperlink, its value is used as the prompt input for the target report. You can create documents using the link object as you would with any other object. Users can then click the hyperlinks to access more detailed documents related to the link object. To create a link object, use the openDocument function in the objects Select statement. The Select statement for a link object follows this order:
'<a href="http://<servername>:<port>/OpenDocument/<platformSpe cific>?sDocName=<document name>&sType=<document type>&iDo cID=<document id>&lsS<prompt message>='+object SELECT+'">'+ob ject SELECT+'</a>'
The concatenation operator (+) applies for Microsoft Access databases. Use the operator appropriate to your target RDBMS. For more details on the Select statement, creating link objects, and using link objects in InfoView, refer to the Designers Guide.
Sending information to the Viewer analytic in a dashboard using openAnalytic

Using the openAnalytic syntax, you can send information from the following documents saved on the Central Management Server (CMS) to a Viewer analytic on a dashboard: Web Intelligence reports Crystal Reports Text analytic Desktop Intelligence reports Web page analytic Sample syntax:
<a href='http://[WEB APPLICATION LOCATION] aa-open-inlist.jsp?url=openAnalytic.jsp? DocumentName=text+analytic&RepositoryType=C &RepositoryName=PM&DocumentExt=afd &DocumentId=AQ74dXHxW59HlHvZivNsypI&mode=full ' target='hiddenFrame'>[LINK NAME IN DASHBOARD]</a>
The key items in the sample are aa-open-inlist.jsp and target='hid denframe'. 1. In a text editor, copy and paste the sample syntax.
100
2. Replace [WEB APPLICATION LOCATION] with the web application location after http://. For example: http://<server>:<port>/PerformanceManagement/jsp 3. Enter the correct information after the variables. 4. Replace [LINK NAME IN DASHBOARD] with the name of the link as you want it to appear. 5. Copy and paste the text in the following location: For a Web Intelligence, Crystal Report or Desktop Intelligence report, copy the text in the cell of the query and then run a report. For a Text analytic, go to the edit mode and paste the text in the "Enter text or HTML to display" text box and select html. Save the analytic in the public folders. For a Web page analytic, go to the edit mode and in the "Content" tab paste the text in the "Enter the web page URL" text box and click OK. In the "Layout" tab, select from the "Display as" section A link, so only the name is visible. 6. In InfoView, go to Dashboards > Create Corporate Dashboard. 7. Enter the title, select a public folder, and click OK. 8. Click Dashboards > Organize Corporate Dashboards, and select your dashboard. 9. Click Edit Dashboard. 10. From the "Analytic Toolbox", expand the "Corporate Analytics" list to find your report, and drag it to the dashboard layout. 11. From the "Analytic Toolbox", expand the "New Analytic" list, and drag the "Viewer" analytic from the list to the dashboard layout. 12. Click Save, then Exit Edit Mode. Click on any linked item in the report and the destination appears in the viewer.
What is a metric forecaster detrended mean?

The Metric forecaster analytic uses the following detrending algorithm encoding:
101
Predictive analytics Model Gains Chart analytic based on predictive models
Adjacent-point differencing: Double-differencing: Linear regression: Generalized regression on time:
xt = xt-1 xt = xt-1 + (xt-1 xt-2) = 2*xt-1 xt-2 xt = A + B*t xt = A + B*t + C*t + D*log(t) + E*sqrt(t)
The encoded value is the detrended mean of the signal across all data points with the same sequence number. For each detrending algorithm: A prediction is made across all values with sequence=1. Residuals are calculated as (prediction actual). The mean of these residuals across all values with sequence = 1 is used as the cyclicality encoding for points corresponding to sequence=1. The same is applied to data points with other sequence numbers.
Cyclicality encoding on the detrended signal include: Standard seasonality variables (for example, day of week or month of year) Repetitive sequence encoding of adjacent points in time
Model Gains Chart analytic based on predictive models

The Model gains chart analytic: profiles the combined strength of multiple variables for predicting a binary goal metric in terms of cumulative lift. depicts the relationship between the combination of ALL influencer variables in a predictive model and a binary outcome or goal, using a standard cumulative lift graph (used in database marketing).
Model gains charts shows: bins of the model score variable placed in descending order of response rate along the horizontal axis. a model line that shows the cumulative percent of signals achievable by targeting an increasing percent of the total population.
102
a signal or response assumed to be the least frequent value.
You use model gains chart when you want to: assess how much predictive power is contained in the sum total of a suite of variables. determine the best cut-point to use for targeting based on values of a model score variable (when used in conjunction with the Individual List analytic.
Related Topics
Individual List based on predictive models on page 43
Configuring the Model Gains Chart analytic

1. Select a predictive model. 2. Select items from the "Associated influencers" and "Selected Goal" lists using Ctrl + mouse click.
Related Topics
Selecting a predictive model to analyze in a Predictive analytic on page 40
Model Gains Chart tips

The following tips are available to help you configure a model gains chart: Insight Scoring Sample size Leverage sets Regenerate on refresh Cross-sell analysis
Insight
To optimize model gains chart for insight:
103
Include as many potentially relevant variables as possible. Use multi-goal models and Key influencers analytic configurations.
Related Topics
Key Influencers analytic based on predictive models on page 48
Scoring
To optimize model gains chart for scoring: Minimize the number of influencers by eliminating influencers with low net relevance. Avoid multi-goal models.
Sample size
When selecting sample size: Use absolute sample size, not sampling %, for reliable trending. Only use large samples to trend weak (for example low importance) relationships accurately. Use single influencer models or the Variable Profile Box Plot analytic to quantify goal values by bin against 100% samples of large populations. Increase sample size when regeneration rules trigger too often.
Related Topics
Variable Profile Box Plot based on predictive models on page 105
Leverage sets
Use leverage sets, as needed, to increase performance: Restrict population definition to set/subset conditions Use set/subset membership variables
Regenerate on refresh
Use "Regenerate on Refresh" sparingly, because model regeneration takes much longer than a simple refresh action.
104
Predictive analytics Variable Profile Box Plot based on predictive models
Cross-sell analysis
To optimize model gains chart for cross-sell analysis, use the same variables as influencers & goals in multi-goal models.
Note: Not useful for prediction, which requires known goal values.
Variable Profile Box Plot based on predictive models

The Variable profile box plot analytic profiles the relationship between a variable (character or numeric) and a numeric measure using a standard box plot graph. For each subrange or "bin" of the selected variable, a box plot is displayed that depicts the minimum, maximum, and quartiles (25th, 50th and 75th percentiles) of a profile measure's distribution. An alternative configuration displays box plots based on mean and standard deviations. You use variable profile box plot to: understand how the distribution of a profile measure varies by variable bin, for example range, variability and lopsidedness or "skew." By contrast, the Influencer detail analytic is used for profiling variables against multiple measures side-by-side based on a simple aggregation such as mean. analyze how the range and distribution of a Key Performance Indicator (KPI) varies for different values of a focal attribute to understand and improve consistency of performance. summarize the relationship between a focal attribute and profile or outcome measure in a manner that teases out the influence of extreme values and lopsided distributions.
With variable profile box plot, you can answer questions like: How is "sum of transaction fees" distributed for accounts of different ages? What is the distribution of account representative performance by region?
Related Topics
Influencer Detail analytic based on predictive models on page 44
105
Predictive analytics Variable Profile Box Plot based on predictive models
Configuring the Variable Profile Box Plot analytic

1. Select the universe on which the population exists. Click the refresh icon next to "Selected Population" to be sure that you have the latest list. 2. Select a population. 3. Select the influencer and binning. For information on creating a binning, see the Predictive Analysis documentation. 4. Select a metric to use in the plot.
Note: The metric must contain a binning.
5. Select the display options. The statistics are selected by choosing from a standard configuration. The following parameters are the default standard configurations.
Option Description
Selected Vari- Select either Range or Deviation. ability Range: minimum, maximum, 25%, 50%, 75% Deviation: mean, mean + -sigma, mean + -3*sigma Min Low Mid High Max For "Range", enter the minimum value of measure for each bin. For "Deviation", enter -3 sigma. For "Range", enter the 25% percentile for each bin. For "Deviation", enter -1 sigma. For "Range", enter the median value for each bin. For "Deviation", enter the mean. For "Range", enter the 75% percentile for each bin. For "Deviation", enter +1 sigma. For "Range", enter the maximum value of measure for each bin. For "Deviation", enter +3 sigma.
6. Click OK.
106
Predictive analytics Predictive Analysis Terms
Predictive Analysis Terms

What is binning?
In Predictive Analysis, binning allows you to compress the range of values of a variable into a smaller number, for example, binning of age into age groupings. Binning can be explicitly defined or statistically derived.
Related Topics
What is a variable? on page 123
What is a Boolean?
In Predictive Analysis, Boolean is a data type that represents a condition whose values can only be the logical values of TRUE or FALSE. Internally, these are represented as 1 and 0 respectively. Boolean variables are used when a binary goal is desired. Currently only Boolean variables as goals can be profiled in the Model Gains and Influencer Gains analytics. Logical operators and the isinpopulation() function are especially useful for defining Boolean variables.
What is a derived variable?

Derived variables are user-defined data elements that are derived from universe objects and/or set membership. The universe itself already supports variable definitions in the form of measures and dimensions calculated from source data. Application architects can expand the range of variables available for Predictive Analysis in a safer and more convenient way than by changing or adding universes to the installation.
107
In addition, some of the Predictive Analysis calculations require Boolean goal variables, which can only be enabled through the use of derived variables.
Related Topics
What is a Boolean? on page 116 Predictive Analysis model data types on page 119 What is a variable? on page 123
What is a goal?
A goal measures the achievement of a given metric. Goals are either numeric or Boolean and can be positive or negative, depending on the desired outcome. It consists of a target value and, optionally, a tolerance zone. A goal is assigned a goal status based on the performance of its associated metric. When the metric actual value over-performs the target value beyond tolerance, then the goal status is green. If the metric actual value falls within the tolerance range, then the goal status is yellow. If the metric actual value under-performs the target value beyond the tolerance, then the goal status is red.
The status of a goal depends on the trend color of its associated metric: a Revenue metric that is above tolerance has a green status whereas a Cost metric that is above tolerance has a red status. A goal is based on a metric: A goal has the same calendar as the metric it is associated with. For example, a goal based on a monthly metric follows monthly intervals and a goal based on a fiscal week metric follows fiscal week intervals.
108
A goal has the same type as its metric. For example, a goal based on a Revenue metric is typed as Increase is Good.
A metric can have multiple goals associated with it. For example, a metric can have plan, forecast and benchmark goals.
Related Topics
What is goal status? on page 118 What is a Boolean? on page 116
What is goal status?

The goal status helps interpret the performance of a metric in a simple, visual way. The position of the metric actual value within a tolerance zone of a specified period determines the goal status for a given period. Goal status takes into account the metric's polarity. The following colors represent goal status: Green represents a goal that is meeting a target. Yellow represents a goal that is within the tolerance range. Red represents a goal that is below the set target. Gray represents an undetermined status -- either the tolerance or metric value or both values are missing. A goal can also be grey when the metric last period is ahead of the goal last period. The equivalent of the graphical goal status icon exists in the rule engine as a numerical status.
Related Topics
What is a goal? on page 117
What is goal-based binning?

In Predictive Analysis, goal-based binning is optimized for a goal or outcome. Goal-based binnings emphasize distinctions in the value range for an influencer that correspond to significant shifts in the rate or mean value of a specified goal variable.
109
For example, the only important distinctions in age that affect response rates to a marketing campaign are 'under 20, 20-63, and 64+.' The influencer modeling engine automatically calculates these during model generation.
Related Topics
What is binning? on page 116 What is a goal? on page 117 What is a variable? on page 123
What is a grain?
A grain is the duration of the time periods or slices that make up metrics. The metric grain can vary in span size, and is dependent on the calendar; the finer the grain, the shorter the time period between metric value calculations.
Related Topics
What is a metric? on page 121
What is an influencer variable?

An influencer variable is used as a descriptive or behavioral piece of information, such as age, state of residence, or calculations. The primary types of influencer model data types are: Continuous Nominal Ordinal
Related Topics
What is a variable? on page 123 Predictive Analysis model data types on page 119

110
Sources Desktop Intelligence Objects Numeric Date Derived Variable Character Numeric Date
Source data type
Influence data type Nominal
Goal data type
Character
N/A
Continuous, OrdiContinuous, N/A nal N/A Boolean Nominal N/A Nominal N/A Boolean
Nominal data type
Ordinal data type
Related Topics
111
What is a joiner?
In the analysis of sets, a joiner is an individual who was present in a set in a given period but was not present in the prior period. The period of analysis here corresponds to the set refresh period. No metrics are required for this analysis to run.
Related Topics
What is a leaver?
A leaver is an individual who was not present in a set in a given period but was present in the prior period.
What is a metric?
A metric, or KPI, is the measure of actual performance. A metric is defined using objects in the universe, and is usually made up of the following: a measure a filter a date restriction You can use metrics to track actual performance and compare it to a goal, or use it to project a trend. Metrics can also be created on a set universe. Set metrics allow you to track a measure over time for a particular subject within a set.
What is a population?
In Dashboard and Analytics analyses, a population is a named query that defines a group of interest.
112
What is a predictive model?

A predictive model is a user-specified configuration of the predictive calculation engine. To set up a predictive model, the user selects influencers, goals, and the population within which relationships are quantified.
Related Topics
What is a population? on page 121
What is the root-mean-squared-error?

The root-mean-squared error (RMSE) is often the criterion for assigning an appropriate smoothing constant. Select the smoothing constant that gives the smallest RMSE value.
What is the rules engine?

In the Business Objects Dashboard and Analytics applications, you use the rules engine to identify, analyze, predict, and act on a specific event. The rules engine notifies the users when a process is out of control.
What is a sampling?
A sampling is a group of the population with specified information. Ideally you want to analyze a large sample frequently to get the most accurate estimates and the most up-to-date information. In Process Analysis, larger samples make it easier to detect small shifts and make the control chart more sensitive in detecting changes without increasing the rate of false alarms.
Note: Start the sampling when data is available. If no data exists, the
sampling can take longer than necessary to generate.
113
What is a variable?
In Process Analysis and Predictive Analysis, a variable is a measured characteristic or attribute. It can be "actual data" or data derived using a look-up, aggregation or other calculation. A variable can be: A dimension defined in a BusinessObjects universe A measure Derived variables representing a calculation based on measures, dimensions and/or sets that is defined interactively by a user.
Related Topics
What is a derived variable? on page 116
114
Predictive Analysis Terms
Predictive Analysis Terms What is binning?
What is binning?
In Predictive Analysis, binning allows you to compress the range of values of a variable into a smaller number, for example, binning of age into age groupings. Binning can be explicitly defined or statistically derived.
Related Topics
What is a Boolean?
In Predictive Analysis, Boolean is a data type that represents a condition whose values can only be the logical values of TRUE or FALSE. Internally, these are represented as 1 and 0 respectively. Boolean variables are used when a binary goal is desired. Currently only Boolean variables as goals can be profiled in the Model Gains and Influencer Gains analytics. Logical operators and the isinpopulation() function are especially useful for defining Boolean variables.
What is a derived variable?

Derived variables are user-defined data elements that are derived from universe objects and/or set membership. The universe itself already supports variable definitions in the form of measures and dimensions calculated from source data. Application architects can expand the range of variables available for Predictive Analysis in a safer and more convenient way than by changing or adding universes to the installation. In addition, some of the Predictive Analysis calculations require Boolean goal variables, which can only be enabled through the use of derived variables.
Related Topics
116
Predictive Analysis Terms What is a goal?
Predictive Analysis model data types on page 119 What is a variable? on page 123
What is a goal?
A goal measures the achievement of a given metric. Goals are either numeric or Boolean and can be positive or negative, depending on the desired outcome. It consists of a target value and, optionally, a tolerance zone. A goal is assigned a goal status based on the performance of its associated metric. When the metric actual value over-performs the target value beyond tolerance, then the goal status is green. If the metric actual value falls within the tolerance range, then the goal status is yellow. If the metric actual value under-performs the target value beyond the tolerance, then the goal status is red.
The status of a goal depends on the trend color of its associated metric: a Revenue metric that is above tolerance has a green status whereas a Cost metric that is above tolerance has a red status. A goal is based on a metric: A goal has the same calendar as the metric it is associated with. For example, a goal based on a monthly metric follows monthly intervals and a goal based on a fiscal week metric follows fiscal week intervals. A goal has the same type as its metric. For example, a goal based on a Revenue metric is typed as Increase is Good.
A metric can have multiple goals associated with it. For example, a metric can have plan, forecast and benchmark goals.
117
Predictive Analysis Terms What is goal status?
Related Topics
What is goal status? on page 118 What is a Boolean? on page 116
What is goal status?

The goal status helps interpret the performance of a metric in a simple, visual way. The position of the metric actual value within a tolerance zone of a specified period determines the goal status for a given period. Goal status takes into account the metric's polarity. The following colors represent goal status: Green represents a goal that is meeting a target. Yellow represents a goal that is within the tolerance range. Red represents a goal that is below the set target. Gray represents an undetermined status -- either the tolerance or metric value or both values are missing. A goal can also be grey when the metric last period is ahead of the goal last period. The equivalent of the graphical goal status icon exists in the rule engine as a numerical status.
Related Topics
What is a goal? on page 117
What is goal-based binning?

In Predictive Analysis, goal-based binning is optimized for a goal or outcome. Goal-based binnings emphasize distinctions in the value range for an influencer that correspond to significant shifts in the rate or mean value of a specified goal variable. For example, the only important distinctions in age that affect response rates to a marketing campaign are 'under 20, 20-63, and 64+.' The influencer modeling engine automatically calculates these during model generation.
Related Topics
What is binning? on page 116 What is a goal? on page 117
118
Predictive Analysis Terms What is a grain?
What is a grain?
A grain is the duration of the time periods or slices that make up metrics. The metric grain can vary in span size, and is dependent on the calendar; the finer the grain, the shorter the time period between metric value calculations.
Related Topics
What is an influencer variable?

An influencer variable is used as a descriptive or behavioral piece of information, such as age, state of residence, or calculations. The primary types of influencer model data types are: Continuous Nominal Ordinal
Related Topics
What is a variable? on page 123 Predictive Analysis model data types on page 119

Sources Desktop Intelligence Objects Source data type Influence data type Nominal Goal data type
Character
N/A
119
Predictive Analysis Terms What is an influencer variable?
Sources
Source data type
Influence data type
Goal data type
Numeric Date Derived Variable Character Numeric Date
Continuous, OrdiContinuous, N/A nal N/A Boolean Nominal N/A Nominal N/A Boolean
Nominal data type
Ordinal data type
Related Topics
120
Predictive Analysis Terms What is a joiner?
What is a joiner?
In the analysis of sets, a joiner is an individual who was present in a set in a given period but was not present in the prior period. The period of analysis here corresponds to the set refresh period. No metrics are required for this analysis to run.
Related Topics
What is a leaver?
A leaver is an individual who was not present in a set in a given period but was present in the prior period.
What is a metric?
A metric, or KPI, is the measure of actual performance. A metric is defined using objects in the universe, and is usually made up of the following: a measure a filter a date restriction You can use metrics to track actual performance and compare it to a goal, or use it to project a trend. Metrics can also be created on a set universe. Set metrics allow you to track a measure over time for a particular subject within a set.
What is a population?
In Dashboard and Analytics analyses, a population is a named query that defines a group of interest.
121
Predictive Analysis Terms What is a predictive model?
What is a predictive model?

A predictive model is a user-specified configuration of the predictive calculation engine. To set up a predictive model, the user selects influencers, goals, and the population within which relationships are quantified.
Related Topics
What is a population? on page 121
What is the root-mean-squared-error?

The root-mean-squared error (RMSE) is often the criterion for assigning an appropriate smoothing constant. Select the smoothing constant that gives the smallest RMSE value.
What is the rules engine?

In the Business Objects Dashboard and Analytics applications, you use the rules engine to identify, analyze, predict, and act on a specific event. The rules engine notifies the users when a process is out of control.
What is a sampling?
A sampling is a group of the population with specified information. Ideally you want to analyze a large sample frequently to get the most accurate estimates and the most up-to-date information. In Process Analysis, larger samples make it easier to detect small shifts and make the control chart more sensitive in detecting changes without increasing the rate of false alarms.
Note: Start the sampling when data is available. If no data exists, the
sampling can take longer than necessary to generate.
122
Predictive Analysis Terms What is a variable?
What is a variable?
In Process Analysis and Predictive Analysis, a variable is a measured characteristic or attribute. It can be "actual data" or data derived using a look-up, aggregation or other calculation. A variable can be: A dimension defined in a BusinessObjects universe A measure Derived variables representing a calculation based on measures, dimensions and/or sets that is defined interactively by a user.
Related Topics
What is a derived variable? on page 116
123
Predictive Analysis Terms What is a variable?
124
Get More Help
Get More Help
Online documentation library
Business Objects offers a full documentation set covering all products and their deployment. The online documentation library has the most up-to-date version of the Business Objects product documentation. You can browse the library contents, do full-text searches, read guides on line, and download PDF versions. The library is updated regularly with new content as it becomes available. http://support.businessobjects.com/documentation/product_guides/
Additional developer resources
http://devlibrary.businessobjects.com
Online customer support
The Business Objects Customer Support web site contains information about Customer Support programs and services. It also has links to a wide range of technical information including knowledgebase articles, downloads, and support forums. http://www.businessobjects.com/support/
Looking for the best deployment solution for your company?
Business Objects consultants can accompany you from the initial analysis stage to the delivery of your deployment project. Expertise is available in relational and multidimensional databases, in connectivities, database design tools, customized embedding technology, and more. For more information, contact your local sales office, or contact us at: http://www.businessobjects.com/services/consulting/
Looking for training options?
From traditional classroom learning to targeted e-learning seminars, we can offer a training package to suit your learning needs and preferred learning style. Find more information on the Business Objects Education web site: http://www.businessobjects.com/services/training
126
Get More Help
Send us your feedback
Do you have a suggestion on how we can improve our documentation? Is there something you particularly like or have found useful? Drop us a line, and we will do our best to ensure that your suggestion is included in the next release of our documentation: mailto:documentation@businessobjects.com
Note: If your issue concerns a Business Objects product and not the
documentation, please contact our Customer Support experts. For information about Customer Support visit: http://www.businessobjects.com/support/.
Business Objects product information
For information about the full range of Business Objects products, visit: http://www.businessobjects.com.
127
Get More Help
128
Index
A
analytics binning display options 42, 46 linking from a goal, metric or universe query analytic 54 linking to multiple documents 52 linking using variables 55 linking with nbPrompts 70 derived variables (continued) editing 18 detrended mean 101 display options binning 42, 46
F
fact table aggregates 14 Frequency Histogram analytic OpenAnalytic variables 69
B
binning creating 17 deleting 18 editing 18 Bubble Chart analytic OpenAnalytic variables 69
G
Gauge analytic OpenAnalytic variables 67 Goal analytic OpenAnalytic variance variables 57 Goal-based Influencer Detail analytic 41 binning 42, 46 configuring 42 display options 42, 46 goals publishing 32 publishing schedule 34 publishing with actions 34 reviewing publish settings 36 selecting a role for publishing 32 selecting goal slices for publishing 33 selecting the publishing strategy 32 status 109, 118
C
Curve Fitting analytic OpenAnalytic variables 69
D
data types continuous 21, 110, 119 nominal 21, 110, 119 ordinal 21, 110, 119 derived variables 14 as fact table aggregates 14 binnings 17 creating 15 deleting 18
129
Index
I
Individual List analytic 43 configuring 44 Influencer Detail analytic 44 binning 42, 46 configuring 45 display options 42, 46 Influencer Gains Chart analytic 46 configuring 47 Interactive Metric Trend analytic OpenAnalytic variables 65
K
Key Influencers analytic configuring 50 description 48 OpenAnalytic variables 65
L
Lag Plot analytic OpenAnalytic variables 69 linking openDocument 70 to a document or analytic 54 to multiple documents 52
M
Map analytic OpenAnalytic variables 58 Membership analytic OpenAnalytic variables 63 Metric Forecaster analytic configuring 51 description 50 detrended mean 101 OpenAnalytic variables 65
Metric List analytic OpenAnalytic variables 60 Metric Tree analytic OpenAnalytic variables 62 metrics based on models 26 changing the list view 38 creating model-based metrics 26 publishing 32 publishing schedule 34 publishing with actions 34 purging 31 reviewing publish settings 36 selecting a role for publishing 32 selecting goal slices for publishing 33 selecting the publishing strategy 32 Metrics Overview analytic OpenAnalytic variables 62 Migration analytic OpenAnalytic variables 67 Model Gains Chart analytic 102 configuring 103 optimize for cross-sell analysis 105 optimize for insight 103 optimize for scoring 104 regenerate on refresh 104 sample size 104 tips 103 using leverage sets 104 model-based metrics 26 aggregation functions 28 creating 26 deleting 30 goal levels 30 publishing 31, 32 publishing schedule 34 publishing with actions 34 refreshing 30 reviewing publish settings 36 selecting a role for publishing 32 selecting goal slices for publishing 33 selecting the publishing strategy 32
130
Index
models 20 adding 20 copying 22 data types 21, 110, 119 deleting 23 editing 23 regenerating 23 regenerating statistics 23 setting the refresh 20 most volatile metric analytics OpenAnalytic variables 63
openDocument in analytics 70 in dashboards 70
P
Pareto analytic OpenAnalytic variance variables 57 Pie Chart analytic OpenAnalytic variables 69 populations adding 10 analysis 10 changing the list view 12 editing 11 removing 12 predictive analysis description 8 predictive models selecting 40
N
nbPrompts 70 goal and metric-based analytics 70 Normal Probability Plot analytic OpenAnalytic variables 69
O
OpenAnalytic 54 with the Viewer analytic 100 OpenAnalytic variables data exploration analytics 69 for linking analytics 55 gauges 67 goal variance 57 in goal, metric and universe query analytics 55 interactive metric trend analytic 65 key influencers analytic 65 map analytics 58 metric forecaster 65 metric list 60 metric overview 62 metric tree 62 migration analytic 67 pie, bubble and radar charts 69 speedometer 67 strategy map 60 volatile metrics and set membership 63
R
Radar Chart analytic OpenAnalytic variables 69
S
Smoothing analytic OpenAnalytic variables 69 speedometer gauge openanalytic variables 67 statistics for models 23 Strategy Map analytic OpenAnalytic variables 60
U
URL reporting openDocument 71
131
Index
V
Variable Profile Box Plot analytic 105 configuring 106
Viewer analytic having information sent to 100 using openAnalytic 100
132

Xi3 Predictive Analysis User Guide en

Загружено:

Сведения о документе

Исходное описание:

Оригинальное название

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

Xi3 Predictive Analysis User Guide en

Загружено:

Авторское право:

Доступные форматы

BusinessObjects Predictive Analysis User Guide

BusinessObjects Predictive Analysis 3.0 Windows

BusinessObjects Predictive Analysis User Guide

Predictive Analysis model-based metrics

BusinessObjects Predictive Analysis User Guide

BusinessObjects Predictive Analysis User Guide

BusinessObjects Predictive Analysis User Guide

BusinessObjects Predictive Analysis User Guide

Predictive Analysis populations

Predictive Analysis populations Adding a population in Predictive Analysis

Adding a population in Predictive Analysis

BusinessObjects Predictive Analysis User Guide

Predictive Analysis populations Editing a population in Predictive Analysis

Editing a population in Predictive Analysis

Adding a population in Predictive Analysis on page 10

BusinessObjects Predictive Analysis User Guide

Predictive Analysis populations Removing a population in Predictive Analysis

Removing a population in Predictive Analysis

The Populations list in Predictive Analysis

BusinessObjects Predictive Analysis User Guide

Derived Variables in Predictive Analysis

Derived Variables in Predictive Analysis Derived variable as fact table aggregates

Derived variable as fact table aggregates

BusinessObjects Predictive Analysis User Guide

Derived Variables in Predictive Analysis Creating derived variables in Predictive Analysis

Derived Variable Syntax for Measures

Creating derived variables in Predictive Analysis

BusinessObjects Predictive Analysis User Guide

Derived Variables in Predictive Analysis Creating derived variables in Predictive Analysis

BusinessObjects Predictive Analysis User Guide

Derived Variables in Predictive Analysis Creating a binning in Predictive Analysis

Creating a binning in Predictive Analysis

BusinessObjects Predictive Analysis User Guide

11. Click OK.

Editing derived variables and binning in Predictive Analysis

Deleting derived variables and binning in Predictive Analysis

BusinessObjects Predictive Analysis User Guide

Predictive Analysis models

Predictive Analysis models Adding a model in Predictive Analysis

variables are not required.

Adding a model in Predictive Analysis

BusinessObjects Predictive Analysis User Guide

Predictive Analysis models Adding a model in Predictive Analysis

Predictive Analysis model data types

Continuous, OrdiContinuous, N/A nal N/A N/A

BusinessObjects Predictive Analysis User Guide

Predictive Analysis models Copying a model in Predictive Analysis

Sources Derived Variable Character Numeric Date

Source data type Boolean Nominal

Influence data type Nominal N/A

Goal data type Boolean

Continuous, OrdiContinuous, N/A nal N/A N/A

Continuous data type

What is a Boolean? on page 116

Copying a model in Predictive Analysis

BusinessObjects Predictive Analysis User Guide

Predictive Analysis models Editing a model in Predictive Analysis

A copy of the model appears in the list.

Editing a model in Predictive Analysis

Adding a model in Predictive Analysis on page 20

Deleting a model in Predictive Analysis

Regenerating a model in Predictive Analysis