Вы находитесь на странице: 1из 33

Quantitative Methods 2010

Measurement Scales Measures Of Central Tendency Measures Of Dispersion


Quant Methods 2010 - Kingston 1

Overview
The 4 scales of measurement Objectives of averaging Requisites of a good average Types of averages Mathematical and positional averages Range, standard deviation and variance

Quant Methods 2010 - Kingston

Scales Of Measurement

Quant Methods 2010 - Kingston

Referigerator Models In HILL CREST Society Flat No. 101 102 103 104 201 202 203 204 301 302 303 304 401 402 403 404 501 502 503 504 Refrigerator LG Voltas LG LG Samsung Voltas LG Godrej Godrej LG Voltas Voltas Samsung LG LG Godrej Godrej Voltas LG Voltas

Passing Class Obtained In B.Sc. Physics Seat No. 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 B.Sc. Physics Pass Pass Second Pass Pass First Second Pass Pass First Second Pass Pass Second Pass First Pass Pass Second Second

Melting Temperature Of Solder, C Sample No Temp 1 432 2 445 3 429 4 433 5 442 6 443 7 447 8 441 9 448 10 445 11 443 12 443 13 443 14 446 15 447 16 441 17 439 18 440 19 447 20 442

Cholesterol Level Of Managers Emp. No. mg/dL 2660 178.5 2661 152 2662 149.3 2663 164 2664 155 2665 138.5 2666 149.3 2667 146.5 2668 164.2 2669 178.8 2670 149.3 2671 138.2 2672 155.1 2673 146.5 2674 148.2 2675 167 2676 158.1 2677 143.2 2678 139.2 2679 143.3
4

Quant Methods 2010 - Kingston

Four Types Of Scales


Nominal (also called categorical)
Measurement says which category observation falls in Example: Refrigerators data (previous slide) Central tendency: mode Neither mean nor median cannot be defined

Ordinal
Measurement says what is rank of an observation compared to others. Example: B.Sc results Central tendency: mode or median. Mean cannot be defined

Interval
Measurement is position of observation on a scale with arbitrary zero. Example: Melting temperature of solder Central tendency: mode, median or arithmetic mean

Ratio
Measurement is position of an observation on a scale with real zero. Example: Cholesterol level of managers Central tendency: mode, median or arithmetic mean

Quant Methods 2010 - Kingston

Measures of Central Tendency

Quant Methods 2010 - Kingston

Objectives Of Averaging
To get a single value that in some sense is representative of the individual data-points.
To facilitate comparisons between data-sets.

Quant Methods 2010 - Kingston

Requisites Of A Good Average


Should be
Well defined Easy to compute Capable of simple interpretation Dependent on all observed values Not influenced much by outliers Similar from one random sample to next Capable of mathematical manipulation
Quant Methods 2010 - Kingston 8

Types Of Averages
Averages

Mathematical

Positional

Arithmetic Mean

Geometric Mean

Harmonic Mean

Median

Mode

Quant Methods 2010 - Kingston

Arithmetic Mean Continuous, Ungrouped Data


If data is individual continuous observations called xi , mean is written: x

x = (sum of all x) / total no of x

= xi / n i 1
Company Name Ashok Leyland Classic Finance Empire Finance First Leasing Llyods Finance Profit, Rs Crore 9.19 4.27 1.74 5.71 4.8 Total 25.71 Arith Mean (average) 5.142
Quant Methods 2010 - Kingston 10

Arithmetic Mean Discrete, Ungrouped Data


If data is discrete values, called xi , with frequencies (counts) fi :
x = (sum of all frequency-value cross-products) / (sum of all frequencies)
x

fx
i 1 n i

f
i 1

Profit, Rs. Crores No Of Companies 16 15 20 12 24 8 25 7 31 8 Total 50

Cross-Prod 240 240 192 175 248 1095 Arith Mean:

21.9
11

Quant Methods 2010 - Kingston

Arithmetic Mean Grouped Data


If data is continuous values called xi , grouped into classes, with mid-points mi and frequencies fi :
= (sum of all freq-midpoints cross-prods) / (sum of all frequencies)
x

i 1 n

f i mi
i

f
i 1

Dividend, % No Of Companies 0-8 12 8 - 16 20 16 - 24 25 24 - 32 28 32 - 40 15 Total 100 Average


Quant Methods 2010 - Kingston

Midpoint 4 12 20 28 36

Cross-Prod 48 240 500 784 540 2112 21.12


12

Weighted Average
If individual values are not to be treated as equally important, but weights are assigned (according to some rationale), we can compute weighted average
x

w x
i 1 n i

w
i 1

Company Name Ashok Leyland Classic Finance Empire Finance First Leasing Llyods Finance

Profit, Rs Crore 9.19 4.27 1.74 5.71 4.8 Total 25.71 Arith Mean (average) 5.142

Weightage 5 3 5 4 1 18

Cross-Prod 45.95 12.81 8.7 22.84 4.8 95.1 5.28

Quant Methods 2010 - Kingston

13

Median
Middle value of a sorted series of data Positional average Value below which half the observations fall. If there are odd number of data points, middle value is Median. If there are even number of data points, average of the two central points is Median
Median 23 23 34 34 38 52 52 57 57 38 43
14

Quant Methods 2010 - Kingston

Mode
If data is individual observations, most frequently occurring value is Mode. If data is grouped into frequency distribution, Mode = , where f f1
Lmo

Lmo = lower limit of modal class f1 and f2 = frequencies of preceding and succeeding class respectively f = frequency of modal class W = class interval.
Quant Methods 2010 - Kingston 15

2 f f1 f 2

Examples For Mode


7 4 5 3 4 2 6 4 3 5 Mode
Sales, Crore No Of Co's 0-8 19 8 - 16 25 16 - 24 36 24 - 32 43 32 - 40 28
50 40 30 20 10 Series1

0 0-8 8 - 16 16 - 24 24 - 32 32 - 40

Lmo = 24, f2 = 28, f1 = 36, f = 43, W = 8


Mode = 26.55
Quant Methods 2010 - Kingston 16

Measures Of Dispersion

Quant Methods 2010 - Kingston

17

Range
Difference between highest and lowest data point. Convenient to calculate, but does not give good idea of dispersion because
Many different dispersions can have same range. Outliers give false idea of dispersion.

Quant Methods 2010 - Kingston

18

Shortcomings Of Range
Each of these data-sets have very different dispersion, but same range.

200

1000

200

1000

200

1000

200

1000

Quant Methods 2010 - Kingston

19

Mean Absolute Deviation (MAD)


Deviation of interest is deviation from mean. Sum of deviation of individual data points from Mean (with sign) will be 0. | xi x | MAD =
n

Mean

|x - x| 4 2 0 2 4 6 2.4 Quant Methods 2010 - Kingston

x 2 4 6 8 10

20

Standard Deviation & Variance


Concept of population and sample. Population standard deviation denoted by

(x
i

)2

Sample standard deviation denoted by s.


s

--- is population mean


( xi x ) 2 n 1

--- x is sample mean

Sometimes referred to as root-mean-square (RMS) deviation Squares of standard deviations, 2 and s2 , are called variances (population and sample, respectively).
Quant Methods 2010 - Kingston 21

Example Calculating s From Raw Data


Price Rs/Kg (x) (x - xbar) (x - xbar)^2 1 2 3 4 5 6 7 8 9 10 Ave, xbar Sum s 85.5 91.3 81.9 84.7 83.6 90 88.5 92.2 84.8 86.1 86.86 107.34 3.45
Quant Methods 2010 - Kingston 22

-1.36 4.44 -4.96 -2.16 -3.26 3.14 1.64 5.34 -2.06 -0.76

1.8496 19.7136 24.6016 4.6656 10.6276 9.8596 2.6896 28.5156 4.2436 0.5776

Grouped Data
If data is grouped into classes with midpoints mi , then fm
n

i 1 n

f
i 1

and
s

i 1

f i(mi x) 2

f
i 1

Quant Methods 2010 - Kingston

23

Example Calculating s from Grouped Data


Bin 80-81.9 82-83.9 84-85.9 86-87.9 88-89.9 90-91.9 92-93.9 Frequency Mid-point, m Cross-Prod, fm m - xbar (m - xbar)^2 1 1 3 1 1 2 1 81 83 85 87 89 91 93 81 83 255 87 89 182 93 -6 -4 -2 0 2 4 6 36 16 4 0 4 16 36

Sum f Ave, xbar s

10

870

112 87 3.35

Quant Methods 2010 - Kingston

24

Some Properties Of Std Dev s


s is unaltered if every data point has constant added to it. sx + c = sx
Price Rs/Kg (x) (x - xbar) (x - xbar)^2 1 2 3 4 5 6 7 8 9 10 Ave, xbar Sum s 85.5 91.3 81.9 84.7 83.6 90 88.5 92.2 84.8 86.1 86.86 -1.36 4.44 -4.96 -2.16 -3.26 3.14 1.64 5.34 -2.06 -0.76 1.8496 19.7136 24.6016 4.6656 10.6276 9.8596 2.6896 28.5156 4.2436 0.5776

Price Rs/Kg (x) (x - xbar) 1 2 3 4 5 6 7 8 9 10 90.5 96.3 86.9 89.7 88.6 95 93.5 97.2 89.8 91.1 91.86 -1.36 4.44 -4.96 -2.16 -3.26 3.14 1.64 5.34 -2.06 -0.76

(x - xbar)^2 1.8496 19.7136 24.6016 4.6656 10.6276 9.8596 2.6896 28.5156 4.2436 0.5776

Ave, xbar Sum 107.34 Quant Methods 2010 - Kingston s 3.45

107.34 25 3.45

Some Properties Of Std Dev s


s is unaltered if every data point has a constant multiplied to it : scx = csx Price
Rs/Kg (x) (x - xbar) (x - xbar)^2
Price Rs/Kg (x) (x - xbar) (x - xbar)^2 1 2 3 4 5 6 7 8 9 10 Ave, xbar Sum s 85.5 91.3 81.9 84.7 83.6 90 88.5 92.2 84.8 86.1 86.86 107.34 3.45
Quant Methods 2010 - Kingston

-1.36 4.44 -4.96 -2.16 -3.26 3.14 1.64 5.34 -2.06 -0.76

1.8496 19.7136 24.6016 4.6656 10.6276 9.8596 2.6896 28.5156 4.2436 0.5776

1 2 3 4 5 6 7 8 9 10 Ave, xbar Sum s

171 182.6 163.8 169.4 167.2 180 177 184.4 169.6 172.2 173.72

-2.72 8.88 -9.92 -4.32 -6.52 6.28 3.28 10.68 -4.12 -1.52

7.3984 78.8544 98.4064 18.6624 42.5104 39.4384 10.7584 114.0624 16.9744 2.3104

429.38 6.91
26

Coefficient Of Variation (CV)


An s is good or bad (low or high) depending on its relationship to the mean.
Example: s = 3 would be bad if xbar = 5, but good if xbar = 500

CV = s/ xbar
Can be expressed in percent also ( Mult by 100) CV < 20% considered good.

Quant Methods 2010 - Kingston

27

Try This Yourself


Break-strength, kg 1 2 3 4 5 6 7 8 9 10 55.8 49.3 52.6 57.4 58.8 47.5 52.1 61.7 53.9 56.6
1 Compute the standard deviation of break-strength. 2 Make a new table, adding 5 to each data point. Compute new standard deviation. Verify that new standard deviation is unaltered. 3 Make a new table, multiplying each data point by 2. Compute new standard deviation. Verify that new standard deviation is multiplied by 2. 4 Group data into classes of width 3, and show frequency distribution. Compute standard deviation of grouped data.

Quant Methods 2010 - Kingston

28

Break-strength, kg (x - xbar) (x - xbar)^2 1 2 3 4 5 6 7 8 9 10 Ave, xbar Sum s 55.8 49.3 52.6 57.4 58.8 47.5 52.1 61.7 53.9 56.6 54.57
Quant Methods 2010 - Kingston

Working Of Q 1

1.23 -5.27 -1.97 2.83 4.23 -7.07 -2.47 7.13 -0.67 2.03

1.5129 27.7729 3.8809 8.0089 17.8929 49.9849 6.1009 50.8369 0.4489 4.1209

170.56 4.35

29

Adding 5 to each x Breakstrength, kg (x - xbar) 1 2 3 4 5 6 7 8 9 10 Ave, xbar Sum s 60.8 54.3 57.6 62.4 63.8 52.5 57.1 66.7 58.9 61.6 59.57 1.23 -5.27 -1.97 2.83 4.23 -7.07 -2.47 7.13 -0.67 2.03

Working Of Q 2

(x - xbar)^2 1.5129 27.7729 3.8809 8.0089 17.8929 49.9849 6.1009 50.8369 0.4489 4.1209

Quant Methods 2010 - Kingston

170.56 4.35

30

Multiplying each x by 2 Breakstrength, kg (x - xbar) (x - xbar)^2 1 2 3 4 5 6 7 8 9 10 Ave, xbar Sum s 121.6 108.6 115.2 124.8 127.6 105 114.2 133.4 117.8 123.2 119.14 682.24 8.71
Quant Methods 2010 - Kingston

Working Of Q 3

2.46 -10.54 -3.94 5.66 8.46 -14.14 -4.94 14.26 -1.34 4.06

6.0516 111.0916 15.5236 32.0356 71.5716 199.9396 24.4036 203.3476 1.7956 16.4836

31

Working of Q 4

Bin 47.1- 50 50.1-53 53.1-56 56.1-59 59.1-62 62.1-65

Frequency Mid-point, m Cross-Prod, fm m - xbar (m - xbar)^2 2 2 2 3 1 0 48.5 51.5 54.5 57.5 60.5 97 103 109 172.5 60.5 -5.7 -2.7 0.3 3.3 6.3 32.49 7.29 0.09 10.89 39.69

Sum f Ave, xbar s

10

542

90.45 54.2 3.01

Quant Methods 2010 - Kingston

32

End Of Measurement Scales Central Tendency - Dispersion

Quant Methods 2010 - Kingston

33

Вам также может понравиться