Вы находитесь на странице: 1из 19

Experimental Design of Determining the

Factors That Affect International Flight


Ticket Prices

Nur Dayana Abd Rahman


Industrial Engineering
University of Louisville
07/16/2010
Experimental Design of Determining the Factors That Affect
International Flight Ticket Prices

Nur Dayana Abd Rahman


Industrial Engineering
University of Louisville
07/16/2010

Executive Summary

An experiment was designed to study the factors that affect a round-trip for an economy class
international airline ticket price to travel from the United States of America (USA) to Malaysia.
Three factors were included in the design which was the month to travel to the desired
destination, day of departure from USA to Malaysia and type of airline chosen to fly to Malaysia.
Two different months to travel (December and May), two different departure days (Wednesday
and Friday) and three different airlines (Cathay Pacific, China Southern Airlines and China
Airlines) were chosen for this experiment.

Several statistical procedures were used to determine whether the factors were significantly
contributing to the international airfare. Kruskal-Wallis test, a nonparametric statistical method,
was used to test for the null hypotheses of the experiment. From the results, it can be concluded
that month to travel and the type of airline chosen are the two significant factors that affect the
international airfare.

i
Table of Contents

Executive Summary ....................................................................................................................................... i


Introduction ................................................................................................................................................... 1
Description of Experimental Design ............................................................................................................. 1
Data Analysis & Results ............................................................................................................................... 4
Discussion ..................................................................................................................................................... 9
Conclusions .................................................................................................................................................11
Appendix A: ANOVA Test .........................................................................................................................12
Appendix B: Kruskal-Wallis Test ...............................................................................................................12
Appendix C: Tukey Test .............................................................................................................................15
Appendix D: Regression Analysis ..............................................................................................................15
References ...................................................................................................................................................16

ii
Introduction

Air travel has been known as one of the transportation modes used by mankind since the first
creation of airplane. Traveling by air is a way to reach various destinations whether locally or
internationally, and the purpose of the trip could be for business or leisure purposes. With the
invention of internet search engines, purchasing airline tickets has been made much easier
especially to survey the airline ticket price for the particular date intended for traveling.

However, purchasing an airline ticket online could also be a challenge especially when it comes
to determining the lowest airline fare. Ticket prices fluctuate all the time and there seems to be
several factors that contribute to the fluctuation in the price. A survey done by Seaney (2010)
found that the airline ticket price for domestic North American travel was the cheapest for
Tuesday, Wednesday and Saturday. Nonetheless, Seaney (2010) also noted in the same report
that the cheapest day to travel internationally has been found to be slightly different. Besides
departure day, other factors that were found to significantly affect domestic airline ticket price
are such as flight time, type of airline, travel packages, connecting flight, et cetera.

In this report, an experiment was conducted to study the factors affecting the international airline
ticket price. Time of the year has been found to be a significant factor for traveling (“Discount
Airline Ticket”, September 28, 2003). It was reported in the same website that one of the peak
travel periods is winter holiday time, while period of April through May would be a better time
to get airfares at a discount. The next factor that was considered in the experiment was whether
type of airline flown would affect the international airfare paid for the respective time durations.
Lastly, since departure day was reported to be significant in affecting domestic airfare, this factor
was also included in this experiment.

Description of Experimental Design

The main purpose of this experiment was to determine the cheapest round-trip international
airfare for an economy class from USA to a country in the Southeast Asia. For this experiment,
the destination was chosen to be from USA to Malaysia. Among the factors considered in the
design were time to travel, day of departure and type of airline.

Two different travel times, which are December and May, have been chosen to be included in the
design of the experiment to determine the international airfare. Next, two levels of departure day
were chosen for the experiment and they were Wednesday and Friday. For simplicity, Friday was
fixed as the arrival day at USA for both travel times. The next factor was type of airline and three
types of airline were chosen for the international flight from Los Angeles International Airport,

1
United States of America (USA) to Kuala Lumpur, Malaysia. Cathay Pacific, China Southern
Airlines and China Airlines were the airlines used to investigate if there was any significant
affect on the airfare. In addition, this experiment was conducted in a randomized block design
and day was identified as a blocking factor since ticket price could fluctuate depending on the
day it is bought or viewed online.

This experimental design used three replicates for each level of factor. To gather the data, the
airfare was quoted using three websites which are Expedia.com, Orbitz.com and Vayama.com.
The following dates show the specific dates that were used in quoting the price for a round-trip
ticket of an economy class for an adult from Los Angeles International Airport to Kuala Lumpur
International Airport. It should also be noted that the price was quoted for a one-stop flight.

1. Travel time in December


o Wednesday departure: 12/15/2010-01/07/2011
o Friday departure: 12/17/2010-01/07/2011

2. Travel time in May


o Wednesday departure: 5/4/2011-5/28/2011
o Friday departure: 5/6/2011-5/28/2011

The following three figures show an example of how the data was collected.

Figure 1: Ticket price for flight from Los Angeles International Airport to Kuala Lumpur,
Malaysia from Expedia.com
2
Figure 2: Ticket price for flight from Los Angeles International Airport to Kuala Lumpur,
Malaysia from Orbitz.com

Figure 3: Ticket price for flight from Los Angeles International Airport to Kuala Lumpur,
Malaysia from Vayama.com

3
Data Analysis & Results

Table 1 below shows the price for the international flight from Los Angeles, USA to Kuala
Lumpur, Malaysia for three factors: month of traveling (A), departure day from Los Angeles (B)
and type of airline to fly (C). Data was collected in four different days to block out if there was
any fluctuation in airfare.

Table 1: International airline ticket price collected for four different factors
Month (A) Departure Day (B) Airline ( C) Day 1 Day 2
December Wednesday Cathay Pacific $1,654 $1,943 $1,660 $1,654 $1,943 $1,644
China Southern Airlines $1,670 $1,765 $1,640 $1,670 $1,765 $1,635
China Airlines $1,872 $1,790 $1,846 $1,872 $1,790 $1,846
Friday Cathay Pacific $1,979 $1,604 $1,823 $1,979 $1,604 $1,823
China Southern Airlines $1,775 $1,725 $1,811 $1,775 $1,725 $1,806
China Airlines $1,902 $1,819 $1,876 $1,902 $1,819 $1,876
May Wednesday Cathay Pacific $1,114 $1,114 $1,104 $1,114 $1,114 $1,104
China Southern Airlines $920 $920 $910 $920 $920 $910
China Airlines $1,402 $1,344 $1,381 $1,402 $1,344 $1,381
Friday Cathay Pacific $1,139 $1,139 $1,129 $1,139 $1,139 $1,129
China Southern Airlines $950 $950 $940 $950 $950 $940
China Airlines $1,432 $1,372 $1,410 $1,432 $1,372 $1,410

Month (A) Departure Day (B) Airline ( C) Day 3 Day 4


December Wednesday Cathay Pacific $1,943 $1,943 $1,798 $1,654 $1,943 $1,644
China Southern Airlines $1,671 $1,766 $1,641 $1,671 $1,766 $1,712
China Airlines $1,872 $1,790 $1,846 $1,872 $1,790 $1,846
Friday Cathay Pacific $1,979 $1,604 $1,823 $1,979 $1,979 $1,823
China Southern Airlines $1,851 $1,726 $1,812 $1,776 $1,726 $1,742
China Airlines $1,902 $1,819 $1,876 $1,902 $1,819 $1,876
May Wednesday Cathay Pacific $1,114 $1,114 $1,104 $1,114 $1,114 $1,104
China Southern Airlines $921 $921 $911 $921 $921 $911
China Airlines $1,402 $1,344 $1,381 $1,402 $1,344 $1,381
Friday Cathay Pacific $1,139 $1,139 $1,129 $1,139 $1,139 $1,129
China Southern Airlines $951 $951 $941 $951 $951 $941
China Airlines $1,432 $1,372 $1,410 $1,432 $1,372 $1,410

Minitab software was used to analyze the data for this experiment and an analysis of variance
(ANOVA) test was conducted to test the hypotheses about the equality of the three levels of
treatment means and the interactions between them. Two underlying assumptions are required
when using the ANOVA which are data are normally distributed and data have non-constant
variance. The following equations list the hypotheses tests for ANOVA:

4
    
  
  

(1)

       
(2)

  
  
 
 

(3)

             
(4)

  
  
   
  

(5)
(6)
  
  
   
 (7)

A significance level, α, of 0.05 was used to determine whether to accept or reject the null
hypothesis. Table 2 shows the ANOVA table from Minitab:

Table 2: ANOVA table for experiment on factors affecting international airfare

Source of Variation Sum of Squares (SS) Degrees of Freedom (df) Mean Squares (MS) F-value P-value
Month 15191805 1 15191805 3344.17 0.000
Departure Day 55147 1 55147 12.14 0.001
Airline 1984503 2 1984503 218.42 0.000
Month*Departure Day 4467 1 4467 0.98 0.323
Month*Airline 723336 2 361668 79.61 0.000
Departure Day*Airline 3096 2 1548 0.34 0.712
Month*Departure Day*Airline 2712 2 1356 0.30 0.742
Day 6065 3 2022 0.45 0.721
Error 586018 129 4543
Total 18557149 143

Based on the above table, the p-values for month, departure day, type of airline and the
interaction between month and type of airline were calculated to be less than α = 0.05. Therefore,
the null hypotheses were rejected and these four factors are significant in affecting the airfare.
Next, a model adequacy checking was carried out to check if the ANOVA assumptions were
valid. The following figure, Figure 4, shows the residual plots for the data.

5
Figure 4: Residual plots for ANOVA analysis

From the above figure, the normal probability plot looks good which verified our normality
assumption. However, the residual versus fitted value plot seems to be showing a pattern which
violates the non-constant variance assumption. To approach this problem, several variance-
stabilizing transformations were applied to the response and the residual versus fitted value were
analyzed. The results, however, showed that the normality assumption could not be justified.

Therefore, the ANOVA test is not valid and the Kruskal-Wallis test, which is the nonparametric
equivalent of ANOVA, was used to analyze the data. In order to use the Kruskal-Wallis test,
each observation, yij, was ranked in ascending order. The rank, Rij, is then used to replace the
observation with the smallest observation having rank 1. In the case of ties or observations
having the same value, the average rank was assigned to each of the tied observations. Let Ri● be
the sum of the ranks in the ith treatment. Equation 8 shows the test statistic, H:

    
    !
(8)
 


where ni is the number of observations in the ith treatment. N is the total number of observations,
and S2 is the variance of the ranks, shown in equation 9:

 $%
   
  "  # 

&
(9)


 #

If the ni are reasonably large, H is distributed approximately as '(


Therefore, the null hypothesis is rejected if  ) '( or if p-value is less than the significance
under the null hypothesis.

level.

6
Minitab was used to compute the test statistic for a series of Kruskal-Wallis tests, as shown in
Appendix B. From the results, the p-values for both month and airline were both 0.000 which is
lower than α = 0.05. Therefore, the null hypothesis is rejected and it can be inferred that month
and airline are the significant factors in this experiment.

Once the significant factors have been determined, the main effects plot for the respective factors
were examined to further analyze which factor level would yield the minimum response. Figure
5 shows the main effect plot for month.

Figure 5: Main effects plot for price versus month of traveling

Table 3: Average price for the respective month of traveling from USA to Malaysia

Month Average Price


May $1,148.50
December $1,798.11

Based on Figure 5, the average price for the respective factor level is summarized in table 3.
Therefore, the airfare in May seems to yield the minimum average price in this experiment.
Figure 6 shows the main effects plot for the next significant factor, which was type of airline.

7
Figure 6: Main effects plot for price versus type of airline

Table 4: Average price for the respective airline to travel from USA to Malaysia

Average
Airline Price
China Southern Airlines $1,333.21
Cathay Pacific $1,466.21
China Airlines $1,620.50

Tukey test was used to determine whether there is any significant difference between the average
prices among the three international airlines for this experiment. From the Minitab output shown
in Appendix C, China Southern Airlines is found to be significant with a minimum average price
of $1,333.21.

A regression analysis was also carried out to build a mathematical model relating response to the
significant factors and the interaction between the factors. However, since both significant
factors are qualitative variables, indicator variables were used. Table 5 shows the indicator
variables and their values used when the data was run in Minitab.

Table 5: Indicator variables for month and airline

Month i1 Airline i2 i3
December 0 Cathay Pacific 0 0
May 1 China Southern Airlines 1 0
China Airlines 0 1

8
The regression analysis equation for this experiment is shown below:

*+,-.  /01  2/2,  34,5  2,6  2,,5  554,,6 (9)

However, the regression model adequacy checking seems to be showing a non-constant variance
due to the funnel-like shape of residual versus fitted value plot. Therefore, the regression model
violates the assumption that the errors are approximately normally distributed with constant
variance. Figure 7 shows the residual plots for the regression model.

Figure 7: Residual plots for regression analysis

Discussion

From the ANOVA test, it can be implied that although the data were normally distributed, they
were also showing a non-constant variance. The plot of the residuals looked like an outward-
opening funnel and applying variance-stabilizing functions such as the square root
transformation or the logarithmic transformation did not seem to improve the residual plots.
Hence, the usual parametric statistical method could not be applied since the constant variance
assumption is violated. Therefore, it is crucial to check the model adequacy of ANOVA by
looking at its residual plots. Model adequacy checking is important to avoid making wrong and
misleading conclusions. The Kruskal-Wallis test was then used as an alternative procedure to the
F test ANOVA that would not depend on the normality and constant variance assumptions.

A mathematical model relating the significant factors to the response could not be accomplished
since the regression analysis assumes the errors are normally distributed with constant variance.

9
However, the main effects plot can be used to determine the factor level(s) for the significant
factors that will yield the minimum average international airline ticket.

From the results, it seems that only two factors were significant in affecting the airfare: month to
travel and type of airline chosen. Departure day did not seem to be significant in affecting the
international airfare although it does affect the airfare for domestic flights as noted by Seaney
(2010). An initial assumption that the day data was collected to be a nuisance factor was also
rejected since the Kruskal-Wallis test showed that the p-value was higher than the significance
level. Therefore, there seems to be no difference in what day the airfare was quoted from the
websites.

The month of May appears to show a significantly lower average airfare compared to December
with a difference of $649.61. This result agrees with an earlier assumption that May is among the
non-peak times of the year to travel unlike December, which has two major holidays such as
Christmas and New Year in it.

As for the type of airline chosen, China Southern Airlines is the most significant airline which
yields the minimum airfare compared to Cathay Pacific and China Airlines. As mentioned in the
design stage of this experiment, the prices were quoted based on a one-stop flight. Thus, further
examination on why the airfare for type of airline chosen differs resulted in the duration of transit
between the flights. The transit for China Southern Airlines seems to be the longest with duration
of approximately 11 hours compared to Cathay Pacific and China Airlines.

Although only three factors were analyzed for this experiment, there seem to be numerous
factors that could possibly contribute to the airfare for international flights. For instance, this
experiment considers Malaysia as the destination of traveling. The price could perhaps be
different if another destination such as Europe was chosen. Furthermore, only three factors were
considered in this experiment. A major finding regarding the duration for transit should also be
analyzed in future experiments to determine if longer transits does in fact lead to a lower
international ticket price.

10
Conclusions

Several statistical procedures were used to determine whether the factors were significantly
contributing to the international airfare. Based on the initial analysis using ANOVA, it was
shown that the data for this experiment consist of a non-constant variance. Therefore, the
Kruskal-Wallis test, a nonparametric statistical method, was used to test for the null hypotheses
of the experiment.

From this experimental design, it can be concluded that month to travel and the type of airline
chosen are the two factors that significantly affect the international airfare. Therefore, if a
traveler would like to travel from the United States to Malaysia, he or she would choose to travel
in May and fly with China Southern Airlines to get the cheapest airfare. Day of departure was
determined to be an insignificant factor for international airline ticket. Additionally, blocking
was proved to be unnecessary in this experiment as day for observation does not affect the
airfare.

However, it should be noted that numerous factors and/or their combinations could also affect
the airfare for a round-trip international flight for one person. The factors that should be
considered in future investigations are such as duration of transit between flights, number of
stops for the trip, combination of different airline for a round-trip travel abroad, et cetera. In
addition, the time to quote the ticket price could also pose as a potential nuisance factor
especially if the price is viewed closer to the date of traveling.

11
Appendix A: ANOVA Test

General Linear Model: Price versus Month, Departure, Airline, Day

Factor Type Levels Values


Month fixed 2 December, May
Departure fixed 2 Friday, Wednesday
Airline fixed 3 Cathay Pacific, China Airlines, China Southern
Day fixed 4 1, 2, 3, 4

Analysis of Variance for Price, using Adjusted SS for Tests

Source DF Seq SS Adj SS Adj MS F P


Month 1 15191805 15191805 15191805 3344.17 0.000
Departure 1 55147 55147 55147 12.14 0.001
Airline 2 1984503 1984503 992251 218.42 0.000
Month*Departure 1 4467 4467 4467 0.98 0.323
Month*Airline 2 723336 723336 361668 79.61 0.000
Departure*Airline 2 3096 3096 1548 0.34 0.712
Month*Departure*Airline 2 2712 2712 1356 0.30 0.742
Day 3 6065 6065 2022 0.45 0.721
Error 129 586018 586018 4543
Total 143 18557149

S = 67.4001 R-Sq = 96.84% R-Sq(adj) = 96.50%

Appendix B: Kruskal-Wallis Test

12
Month Departure Airline Day Price Rank Month Departure Airline Day Price Rank
May Wednesday China Southern 1 910 1.5 May Friday China Airlines 4 1,432 70.5
May Wednesday China Southern 2 910 1.5 December Friday Cathay Pacific 1 1,604 74
May Wednesday China Southern 3 911 3.5 December Friday Cathay Pacific 2 1,604 74
May Wednesday China Southern 4 911 3.5 December Friday Cathay Pacific 3 1,604 74
May Wednesday China Southern 1 920 6.5 December Wednesday China Southern 2 1,635 76
May Wednesday China Southern 1 920 6.5 December Wednesday China Southern 1 1,640 77
May Wednesday China Southern 2 920 6.5 December Wednesday China Southern 3 1,641 78
May Wednesday China Southern 2 920 6.5 December Wednesday Cathay Pacific 2 1,644 79.5
May Wednesday China Southern 3 921 10.5 December Wednesday Cathay Pacific 4 1,644 79.5
May Wednesday China Southern 3 921 10.5 December Wednesday Cathay Pacific 1 1,654 82
May Wednesday China Southern 4 921 10.5 December Wednesday Cathay Pacific 2 1,654 82
May Wednesday China Southern 4 921 10.5 December Wednesday Cathay Pacific 4 1,654 82
May Friday China Southern 1 940 13.5 December Wednesday Cathay Pacific 1 1,660 84
May Friday China Southern 2 940 13.5 December Wednesday China Southern 1 1,670 85.5
May Friday China Southern 3 941 15.5 December Wednesday China Southern 2 1,670 85.5
May Friday China Southern 4 941 15.5 December Wednesday China Southern 3 1,671 87.5
May Friday China Southern 1 950 18.5 December Wednesday China Southern 4 1,671 87.5
May Friday China Southern 1 950 18.5 December Wednesday China Southern 4 1,712 89
May Friday China Southern 2 950 18.5 December Friday China Southern 1 1,725 90.5
May Friday China Southern 2 950 18.5 December Friday China Southern 2 1,725 90.5
May Friday China Southern 3 951 22.5 December Friday China Southern 3 1,726 92.5
May Friday China Southern 3 951 22.5 December Friday China Southern 4 1,726 92.5
May Friday China Southern 4 951 22.5 December Friday China Southern 4 1,742 94
May Friday China Southern 4 951 22.5 December Wednesday China Southern 1 1,765 95.5
May Wednesday Cathay Pacific 1 1,104 26.5 December Wednesday China Southern 2 1,765 95.5
May Wednesday Cathay Pacific 2 1,104 26.5 December Wednesday China Southern 3 1,766 97.5
May Wednesday Cathay Pacific 3 1,104 26.5 December Wednesday China Southern 4 1,766 97.5
May Wednesday Cathay Pacific 4 1,104 26.5 December Friday China Southern 1 1,775 99.5
May Wednesday Cathay Pacific 1 1,114 32.5 December Friday China Southern 2 1,775 99.5
May Wednesday Cathay Pacific 1 1,114 32.5 December Friday China Southern 4 1,776 101
May Wednesday Cathay Pacific 2 1,114 32.5 December Wednesday China Airlines 1 1,790 103.5
May Wednesday Cathay Pacific 2 1,114 32.5 December Wednesday China Airlines 2 1,790 103.5
May Wednesday Cathay Pacific 3 1,114 32.5 December Wednesday China Airlines 3 1,790 103.5
May Wednesday Cathay Pacific 3 1,114 32.5 December Wednesday China Airlines 4 1,790 103.5
May Wednesday Cathay Pacific 4 1,114 32.5 December Wednesday Cathay Pacific 3 1,798 106
May Wednesday Cathay Pacific 4 1,114 32.5 December Friday China Southern 2 1,806 107
May Friday Cathay Pacific 1 1,129 38.5 December Friday China Southern 1 1,811 108
May Friday Cathay Pacific 2 1,129 38.5 December Friday China Southern 3 1,812 109
May Friday Cathay Pacific 3 1,129 38.5 December Friday China Airlines 1 1,819 111.5
May Friday Cathay Pacific 4 1,129 38.5 December Friday China Airlines 2 1,819 111.5
May Friday Cathay Pacific 1 1,139 44.5 December Friday China Airlines 3 1,819 111.5
May Friday Cathay Pacific 1 1,139 44.5 December Friday China Airlines 4 1,819 111.5
May Friday Cathay Pacific 2 1,139 44.5 December Friday Cathay Pacific 1 1,823 115.5
May Friday Cathay Pacific 2 1,139 44.5 December Friday Cathay Pacific 2 1,823 115.5
May Friday Cathay Pacific 3 1,139 44.5 December Friday Cathay Pacific 3 1,823 115.5
May Friday Cathay Pacific 3 1,139 44.5 December Friday Cathay Pacific 4 1,823 115.5
May Friday Cathay Pacific 4 1,139 44.5 December Wednesday China Airlines 1 1,846 119.5
May Friday Cathay Pacific 4 1,139 44.5 December Wednesday China Airlines 2 1,846 119.5
May Wednesday China Airlines 1 1,344 50.5 December Wednesday China Airlines 3 1,846 119.5
May Wednesday China Airlines 2 1,344 50.5 December Wednesday China Airlines 4 1,846 119.5
May Wednesday China Airlines 3 1,344 50.5 December Friday China Southern 3 1,851 122
May Wednesday China Airlines 4 1,344 50.5 December Wednesday China Airlines 1 1,872 124.5
May Friday China Airlines 1 1,372 54.5 December Wednesday China Airlines 2 1,872 124.5
May Friday China Airlines 2 1,372 54.5 December Wednesday China Airlines 3 1,872 124.5
May Friday China Airlines 3 1,372 54.5 December Wednesday China Airlines 4 1,872 124.5
May Friday China Airlines 4 1,372 54.5 December Friday China Airlines 1 1,876 128.5
May Wednesday China Airlines 1 1,381 58.5 December Friday China Airlines 2 1,876 128.5
May Wednesday China Airlines 2 1,381 58.5 December Friday China Airlines 3 1,876 128.5
May Wednesday China Airlines 3 1,381 58.5 December Friday China Airlines 4 1,876 128.5
May Wednesday China Airlines 4 1,381 58.5 December Friday China Airlines 1 1,902 132.5
May Wednesday China Airlines 1 1,402 62.5 December Friday China Airlines 2 1,902 132.5
May Wednesday China Airlines 2 1,402 62.5 December Friday China Airlines 3 1,902 132.5
May Wednesday China Airlines 3 1,402 62.5 December Friday China Airlines 4 1,902 132.5
May Wednesday China Airlines 4 1,402 62.5 December Wednesday Cathay Pacific 1 1,943 137
May Friday China Airlines 1 1,410 66.5 December Wednesday Cathay Pacific 2 1,943 137
May Friday China Airlines 2 1,410 66.5 December Wednesday Cathay Pacific 3 1,943 137
May Friday China Airlines 3 1,410 66.5 December Wednesday Cathay Pacific 3 1,943 137
May Friday China Airlines 4 1,410 66.5 December Wednesday Cathay Pacific 4 1,943 137
May Friday China Airlines 1 1,432 70.5 December Friday Cathay Pacific 1 1,979 142
May Friday China Airlines 2 1,432 70.5 13
December Friday Cathay Pacific 2 1,979 142
May Friday China Airlines 3 1,432 70.5 December Friday Cathay Pacific 3 1,979 142
May Friday China Airlines 4 1,432 70.5 December Friday Cathay Pacific 4 1,979 142
December Friday Cathay Pacific 1 1,604 74 December Friday Cathay Pacific 4 1,979 142
Kruskal-Wallis Test: Price versus Month

Kruskal-Wallis Test on Price

Month N Median Ave Rank Z


December 72 1812 108.5 10.36
May 72 1122 36.5 -10.36
Overall 144 72.5

H = 107.26 DF = 1 P = 0.000
H = 107.35 DF = 1 P = 0.000 (adjusted for ties)

Kruskal-Wallis Test: Price versus Departure

Kruskal-Wallis Test on Price

Departure N Median Ave Rank Z


Friday 72 1518 77.6 1.47
Wednesday 72 1519 67.4 -1.47
Overall 144 72.5

H = 2.16 DF = 1 P = 0.141
H = 2.16 DF = 1 P = 0.141 (adjusted for ties)

Kruskal-Wallis Test: Price versus Airline

Kruskal-Wallis Test on Price

Airline N Median Ave Rank Z


Cathay Pacific 48 1372 74.0 0.30
China Airlines 48 1611 90.3 3.61
China Southern 48 1293 53.3 -3.91
Overall 144 72.5

H = 18.93 DF = 2 P = 0.000
H = 18.94 DF = 2 P = 0.000 (adjusted for ties)

Kruskal-Wallis Test: Price versus Day

Kruskal-Wallis Test on Price

Day N Median Ave Rank Z


1 36 1518 71.0 -0.24
2 36 1518 70.9 -0.27
3 36 1518 74.6 0.35
4 36 1538 73.5 0.17
Overall 144 72.5

H = 0.21 DF = 3 P = 0.976
H = 0.21 DF = 3 P = 0.976 (adjusted for ties)

14
Appendix C: Tukey Test
(a) Month

Tukey Simultaneous Tests


Response Variable Price
All Pairwise Comparisons among Levels of Month
Month = December subtracted from:

Difference SE of Adjusted
Month of Means Difference T-Value P-Value
May -649.6 11.23 -57.83 0.0000

(b) Type of airline


Tukey Simultaneous Tests
Response Variable Price
All Pairwise Comparisons among Levels of Airline
Airline = Cathay Pacific subtracted from:

Difference SE of Adjusted
Airline of Means Difference T-Value P-Value
China Airlines 154.3 13.76 11.215 0.0000
China Southern -133.0 13.76 -9.667 0.0000

Airline = China Airlines subtracted from:

Difference SE of Adjusted
Airline of Means Difference T-Value P-Value
China Southern -287.3 13.76 -20.88 0.0000

Appendix D: Regression Analysis

Regression Analysis: Price versus i1, i2, i3, i1*i2, i1*i3

The regression equation is


Price = 1809 - 686 i1 - 75.0 i2 + 41.6 i3 - 116 i1*i2 + 225 i1*i3

Predictor Coef SE Coef T P


Constant 1809.25 14.09 128.41 0.000
i1 -686.08 19.93 -34.43 0.000
i2 -75.00 19.93 -3.76 0.000
i3 41.58 19.93 2.09 0.039
i1*i2 -116.00 28.18 -4.12 0.000
i1*i3 225.42 28.18 8.00 0.000

S = 69.0255 R-Sq = 96.5% R-Sq(adj) = 96.3%

Analysis of Variance

Source DF SS MS F P
Regression 5 17899644 3579929 751.37 0.000
Residual Error 138 657504 4765
Total 143 18557149

15
References

1. Seaney, R. (2010, June 28). Cheapest Day to Fly is Wednesday for U.S. and Canadian
Flights. Retrieved July 14, 2010, from http://www.farecompare.com/articles/airline-
industry-news/cheapest-day-us-canada-flights/
2. Discount Airline Tickets. (2003, September 28). Retrieved July 14, 2010, from
http://brianx.com/travel-guide/discount-airline-tickets.html
3. Montgomery, D.C. (2009). Design and analysis of experiments: International Student
Version. 7th edition. Asia: John Wiley & Sons (Asia) Pte Ltd
4. Orbitz Travel: Airline tickets, cheap hotels, car rentals, vacations & cruises [Data set].
Retrieved from http://www.orbitz.com/
5. Expedia Travel: Airline tickets, hotels, car rental, airfares & vacations [Data set].
Retrieved from http://www.expedia.com/default.asp
6. Vayama: Cheap international airline tickets, international flights, international flight
deals [Data set]. Retrieved from http://www.vayama.com/

16

Вам также может понравиться