B.

Tech II Semester (2020 Batch)

PROBABILITY AND STATISTICS (20BM1104)
(For CSE-3 & CSE-4)

Unit – 5: Correlation and Regression

(The method of least squares, curvilinear regression, multiple regressions, correlation (excluding
causation))

Curve fitting: Computing a curve corresponding to a given set of points is called a curve fitting

Regression: A relation between independent and dependent variables obtained from a given set of points is
called a regression

Simple regression: A relation between one dependent variable and one independent variable obtained
from a given set of points is called a simple regression

Multiple regression: A relation between one dependent variable and two or more independent variables
obtained from a given set of points is called a simple regression

Regression line of y on x : A relation of the form y  a  bx is called a regression line of y on x

Regression line of x on y : A relation of the form x  a  by is called a regression line of x on y

Regression curve of y on x1 , x2 : A relation of the form y  a  bx1  cx2 is called a regression curve of
y on x1 , x2
Least Squares Method: The method of computing a curve (or regression curve) by using a given set of
points such that the sum of the squares of deviations from the points to the curve along y  axis is
minimum

Curve fitting by Least Squares Method:

1. To fit a straight line of the form y  a  bx , the Normal equation are given by
 y  na  b  x
 xy  a  x  b  x 2

2. To fit a straight line of the form x  a  by , the Normal equation are given by
 x  na  b  y
 xy  a  y  b  y 2

3. To fit an exponential curve of the form y  ae bx ,

First write ln y  ln a  bx and then the Normal equation are given by
 ln y  n ln a  b  x
 x ln y  ln a  x  b  x 2
4. To fit an exponential curve of the form y  ab x ,
First write log y  log a  x log b and then the Normal equation are given by
 log y  n log a  log b  x
 x log y  log a  x  log b  x 2

5. To fit a power curve (geometric curve) of the form y  ax b ,

First write log y  log a  b log x and then the Normal equation are given by
 log y  n log a  b  log x
 log x log y  log a  log x  b  (log x) 2

6. To fit a parabola of 2nd degree (or quadratic curve) of the form y  a  bx  cx 2 , the Normal equation
are given by
 y  na  b  x  c  x 2
 xy  a  x  b  x  c  x
2 3

x y  ax  bx  cx

2 2 3 4

7. To fit a multiple regression curve of the form z  a  bx  cy , the Normal equation are given by
 z  na  b  x  c  y
 xz  a  x  b  x  c  xy
2

 yz  a  y  b  xy  c  y 2

8. To fit a multiple regression curve of the form y  a  bx1  cx2 , the Normal equation are given by
 y  na  b  x1  c  x2
x y  ax  bx  cx x
1 1
2
1 1 2

x y  ax  bx x  bx

2 2 1 2
2
2
Problems:

1. Fit a straight line y  a  bx for the following data by least squares method

x 1 2 3 4 5
y 12 25 40 50 65

Solution: The normal equations for the straight line y  a  bx are

 y  na  b x  xy  a x  b x
2
and
Consider
x y x2 xy
1 12 1 12
2 25 4 50
3 40 9 120
4 50 16 200
5 65 25 325
 x  15  y  192  x 2  55  xy  707
Here  x  15,  y  192,  x 2
 55,  xy  707 and n5
The normal equations becomes 192  5a  15b   (1)
and 707  15a  55b   (2)
Solving (1) and (2), a  0.9 and b  13.1
Hence the straight line is y  0.9  13.1x

2. By the method of least squares, fit a straight line y  a  bx for the following data

x 50 70 100 120
y 12 15 21 25

Solution: The normal equations for the straight line y  a  bx are

 y  na  b x  xy  a x  b x
2
and
Consider
x y x2 xy
50 12 2500 600
70 15 4900 1050
100 21 10000 2100
120 25 14400 3000
 x  340  y  73  x 2  31800  xy  6750
Here  x  340,  y  73,  x 2
 31800,  xy  6750 and n  4
The normal equations becomes 73  4a  340 b    (1)
and 6750  340a  31800 b   (2)
Solving (1) and (2), a  2.2758 and b  0.1879
Hence the straight line is y  2.2758  0.1879 x
3. By the method of least squares, fit a straight line x  a  by for the following data

x 12 15 21 25
y 50 70 100 120

Solution: The normal equations for the straight line x  a  by are

 x  na  b y  xy  a y  b y
2
and
Consider
x y y2 xy
12 50 2500 600
15 70 4900 1050
21 100 10000 2100
25 120 14400 3000
 x  73  y  340  y 2  31800  xy  6750
Here  x  73,  y  340,  y 2
 31800,  xy  6750 and n  4
The normal equations becomes 73  4a  340 b    (1)
and 6750  340a  31800 b   (2)
Solving (1) and (2), a  2.2785 and b  0.1879
Hence the straight line is x  2.2785  0.1879 y

4. Fit an exponential curve y  ae bx for the following data

x 1 3 5 7 9
y 100 81 73 54 43

Solution: The curve is y  aebx

Taking log on both sides, log y  log a  bx log e
That is, Y  A  Bx , where Y  log y, A  log a, B  b log e
Now the normal equations for Y  A  Bx are
Y  nA  B x and  xY  A x  B x
2

Consider,
x y Y  log y x2 xY
1 100 2 1 2
3 81 1.9085 9 5.7255
5 73 1.8633 25 9.3165
7 54 1.7324 49 12.1268
9 43 1.6335 81 14.7015
 x  25 Y  9.1377  x  165 2
 xY  43.8703
Here  x  25, Y  9.1377,  x 2
 165,  xY  43.8703 and n  5
The normal equations becomes 9.1377  5 A  25B    (1)
and 43.8703  25 A  165B   (2)
Solving (1) and (2), A  2.0548 and B  0.0455 or
 B
Therefore, a  10 A 113.4488 and b   0.1048
log e
Hence the required curve is y  113.4488 e 0.1048x

5. Fit an exponential curve y  ab x for the following data

x 1 2 3 4 5
y 130 152.2 177.3 190.2 244.7

Solution: The curve is y  ab x

Taking log on both sides, log y  log a  x log b
That is, Y  A  Bx , where Y  log y, A  log a, B  log b
Now the normal equations for Y  A  Bx are
Y  nA  B x and  xY  A x  B x
2

Consider,
x y Y  log y x2 xY
1 130 2.1139 1 2.1139
2 152.2 2.1824 4 4.3648
3 177.3 2.2487 9 6.7461
4 190.2 2.2792 16 9.1168
5 244.7 2.3886 25 11.9432
 x  15 Y  11.2129  x 2  55  xY  34.2849
Here  x  15, Y  11.2129,  x 2
 55,  xY  34.2849 and n  5
The normal equations becomes 11.2129  5 A  15B    (1)
and 34.2849  15 A  55B    (2)
Solving (1) and (2), A  2.0487 and B  0.0646 or
Therefore, a  10 111.8716 and b  10 1.1604
A B

Hence the required curve is y  111.8716 1.1604

x

6. Fit an exponential curve y  ab x for the following data

x 2 3 4 5 6
y 144 172.8 207.4 248.8 298.6

log y  log a  x log b or Y  A  Bx ,where Y  log y, A  log a, B  log b

The normal equations for Y  A  Bx are
Y  nA  B x and  xY  A x  B x
2

Here  x  20, Y   log y  11.5837,  x  90,  xY   x log y  47.1266

2

and n  5
The normal equations becomes 11.5837  5 A  20B and 47.1266  20 A  90B
On solving, A  2 and B  0.0792 or a 100 and b 1.2
Hence the exponential curve y  100 1.2
x
7. Fit a power curve y  ax b for the following data
x 1 2 3 4 5 6
y 2.98 4.26 5.21 6.10 6.80 7.50

Solution: The power curve is y  ax b

Taking log on both sides, log y  log a  b log x
That is, Y  A  bX , where Y  log y, X  log x, A  log a
Now the normal equations for Y  A  bX are
Y  nA  b X and  XY  A X  b X 2
Consider,
x y X  log x Y  log y X2 XY
1 2.98 0 0.4742 0 0
2 4.26 0.3010 0.6294 0.0906 0.1895
3 5.21 0.4771 0.7168 0.2276 0.3420
4 6.10 0.6021 0.7853 0.3625 0.4728
5 6.80 0.6990 0.8325 0.4886 0.5819
6 7.50 0.7782 0.8751 0.6055 0.6809
 X  2.8573 Y  4.3134  X 2  1.7748  XY  2.2671
Here  X  2.8573, Y  4.3134  X  1.7748,  XY  2.2671and n  6
2

The normal equations become 4.3134  6 A  2.8573 b    (1)

and 2.2671  2.8573 A  1.7748 b   (2)
Solving (1) and (2), A  0.4740 and b  0.5143
Therefore, a  10 A  2.9783
Hence the curve is y  2.9783 x 
0.5143

8. Determine the regression line of y on x for the following data

x 20 25 28 35 43
y 52 48 63 79 95

Solution: The regression line of y on x is given by y  a  bx

The normal equations for are
 y  na  b x and  xy  a x  b x
2

Consider
x y x2 xy
20 52 400 1040
25 48 625 1200
28 63 784 1764
35 79 1225 2765
43 95 1849 4085
 x  151  y  337  x 2  4883  xy  10854
Here  x  151,  y  337,  x 2
 4883,  xy  10854 and n5
The normal equations become 337  5a  151b   (1)
and 10854  151a  4883b   (2)
 Solving (1) and (2), a  4.0998 and b  2.096
Hence the regression line of y on x is y  4.0998  2.096 x

9. Determine the regression line of x on y for the following data

x 1.1 2.3 4.5 7.6

y 21 35 64 84
And estimate the value of x at y  4.8

Solution: The regression line of x on y is given by x  a  by

The normal equations for the straight line x  a  by are
 x  na  b y  xy  a y  b y
2
and
Consider
x y y2 xy
1.1 21 441 23.1
2.3 35 1225 80.5
4.5 64 4096 288
7.6 84 7056 638.4
 x  15.5  y  204  y 2  12818  xy  1030
Here  x  15.5,  y  204,  y 2
 12818,  xy  1030 and n  4
The normal equations become 15.5  4a  204 b    (1)
and 1030  204a  12818 b   (2)
Solving (1) and (2), a  1.1849 and b  0.0992
Hence the regression line of x on y is x  1.1849  0.0992 y
Therefore, the value of x at y  4.8 is given by x  1.1849  0.0992 (4.8)  0.70874

10. Fit a second degree parabola y  a  bx  cx 2 for the following data

x 1 3 5 7 9
y 2 7 10 11 9

Solution: The curve is y  a  bx  cx 2

The normal equations are
 y  na  b x  c x 2 ,
 xy  a x  b x  c x
2 3

 x y  a  x  b x  c  x
2 2 3 4

Consider,
x y x2 x3 x4 xy x2 y
1 2 1 1 1 2 2
3 7 9 27 81 21 63
5 10 25 125 625 50 250
7 11 49 343 2401 77 539
9 9 81 729 6561 81 729
 x  25  y  39  x  165
2
 x  1225
3
 x 4  9669  xy  231  x y  1583
2
 Here  x  25,  y  39,  x 2
 165, x 3
 1225, x 4
 9669,  xy  231
x 2
y  1583 and n  5

The normal equations 39  5 a  25b  165c    (1)

231  25 a  165 b  1225 c   (2)
and 1583  165 a  1225 b  9669 c   (3)
Solving (1), (2) and (3), a  1.5571, b  3.7571 and c  0.2857
Hence the parabola is y  1.5571  3.7571x  0.2857 x 2

11. Fit a second degree parabola y  a  bx  cx 2 for the following data

x 1.0 1.5 2.0 2.5 3.0 3.5 4.0

y 1.1 1.3 1.6 2.0 2.7 3.4 4.1

Solution: The curve is y  a  bx  cx 2

The normal equations are
 y  na  b x  c x 2 ,
 xy  a x  b x  c x 2 3

 x y  a  x  b x  c  x
2 2 3 4

Consider,
x y x2 x3 x4 xy x2 y
1.0 1.1 1 1 1 1.1 1.1
1.5 1.3 2.25 3.375 5.0625 1.95 2.925
2.0 1.6 4 8 16 3.2 6.4
2.5 2.0 6.25 15.625 39.0625 5 12.5
3.0 2.7 9 27 81 8.1 24.3
3.5 3.4 12.25 42.875 150.0625 11.9 41.65
4.0 4.1 16 64 256 0.4 1.6
x  y  x2   x3   x4   xy   x2 y 
17.5 12.2 50.75 161.875 548.1875 31.65 90.475

The normal equations 12.2  7 a  17.5b  50.75c    (1)

31.65  17.5 a  50.75 b  161.875 c   (2)
and 90.475  50.75 a  161.875 b  548.1875 c   (3)
Solving (1), (2) and (3), a  2.3929, b  3.7119 and c  0.7095
Hence the parabola is y  2.3929  3.7119 x  0.7095 x 2
12. Determine the Normal equations to fit a straight line of the form y  a  bx
Solution: Let y  f (x) , where f ( x)  a  bx
Let A  ( xi yi ) be any given data point

At x  xi , the observed (given) value of y is yi ; that is AC  yi

At x  xi , the expected value of y is f ( xi ) ; that is BC  f ( xi )
Therefore, the deviation (d i ) at x  xi is given by AB  AC  BC
That is, d i  yi  f ( xi ) so that di2  yi  f ( xi )
2

Sum of the squares of the deviations is given by S   di2   yi  f ( xi )   yi  (a  bxi )
2 2

According least squares method S is minimum

S S
To get S minimum, we need  0 and 0
a b
S
Now  0  2  yi  (a  bxi )(1)  0
a
   yi  (a  bxi )  0
  yi   a   bxi
  yi  n a  b  xi    (1)
S
And  0  2  yi  (a  bxi )( xi )  0
b
 
  xi yi  (axi  bxi2 )  0
  xi yi   axi   bxi2
  xi yi  a  xi  b  xi2    (2)
Therefore (1) and (2) are the required normal equations
13. Determine the least squares regression equation of the form y  a  bx1  cx2 for the following data

y 3 5 6 8 12 14
x1 16 10 7 4 3 2
x2 90 72 54 42 30 12

Solution: The equation is y  a  bx1  cx2

The normal equations are
 y  na  b x1  c x2 ,
 x y  a x  b x  c x x
1 1
2
1 1 2

 x y  a x  b x x  c x
2 2 1 2
2
2

Consider,
y x1 x2 x12 x 22 x1 x2 x1 y x2 y
3 16 90 256 8100 1440 48 270
5 10 72 100 5184 720 50 360
6 7 54 49 2916 378 42 324
8 4 42 16 1764 168 32 336
12 3 30 9 900 90 36 360
14 2 12 4 144 24 28 168
y   x1   x2   x12   x22   x1 x2   x1 y   x2 y 
48 42 300 434 19008 2820 236 1818

The normal equations 48  6 a  42 b  300 c    (1)

236  42 a  434 b  2820 c   (2)
and 1818  300 a  2820 b  19008 c   (3)
Solving (1), (2) and (3), a  16.1067, b  0.4270 and c  0.2219
Hence the regression equation is y  16.1067  0.4270 x1  0.2219 x2

14. Determine the least squares regression equation of the form z  a  bx  cy for the following data

z 16 19 23 20 26 23 28
x 1 2 3 4 5 6 7
y 4 5 7 2 6 1 4

Solution: The equation is z  a  bx  cy

The normal equations are
 z  na  b x  c y ,
 xz  a x  b x  c xy
2

 yz  a y  b xy  c y 2
 Consider,
z x y x2 y2 xy xz yz
16 1 4 1 16 4 16 64
19 2 5 4 25 10 38 95
23 3 7 9 49 21 69 161
20 4 2 16 4 8 80 40
26 5 6 25 36 30 130 156
23 6 1 36 1 6 138 23
28 7 4 49 16 28 196 112
z  x  y   x2   y2   xy   xz   yz 
155 28 29 140 147 107 667 651

The normal equations 155  7 a  28 b  29 c    (1)

667  28 a  140 b  107 c   (2)
and 651  29 a  107 b  147 c   (3)
Solving (1), (2) and (3), a  10, b  2 and c  1
Hence the regression equation is z  10  2 x  y

Exercise:

1. Fit a straight line y  a  bx for the following data by least squares method

x 1 2 3 4 5 6
y 14 33 40 63 76 85

2. Fit a straight line y  a  bx for the following data by least squares method

x 0 2 3 5 9
y -3 4 3 8 15

3. Fit a straight line x  a  by for the following data by least squares method

x 18.5 25.4 30 64.5 34.6 89.8 20.8

y 5 8 10 25 12 36 6
(Ans: x  7  2.3 y )

4. The following shows the improvement of eight students in a speed-reading program, and the number
of weeks they have been in the program:

No.of weeks x 3 5 2 8 6 9 3 4
Speed gain (words/min.) y 86 118 49 193 164 232 73 109

Fit a straight line by the method of least squares

5. If p is the pull required to lift a load w by means of a pulley block, find a linear law of the form
p  mw  c using the data
p 12 15 21 25
w 50 70 100 120

Predict y at x  3.75 by fitting power curve y  a x for the following data

b
6.

x 1 2 3 4 5 6
y 2.98 4.26 5.21 6.10 6.80 7.50

7. Fit a second degree parabola y  a  bx  cx 2 for the following data

x 2.5 3.6 4.6 5.2 6.8 7.2 8.9 9.2

y 1.8 2.6 4.8 6.2 8.9 4.2 2.9 4.5

(Ans: y  7.7504  4.422 x  0.3472 x 2 )

8. Determine the regression line of y on x for the following data

x 50 60 70 90 100
y 65 51 40 26 08

9. Find Y when X1 = 10 and X2 = 6 from the least square regression equation of Y on X1 and X2for the
following data
Y 90 72 54 42 30 12
X1 3 5 6 8 12 14
X2 16 10 7 4 3 2

10. Fit a least-squares regression plane for the following data and also find y at x1  2.2 and x2  90 .

y 5.3 7.8 7.4 9.8 10.8 9.1 8.1 7.2 6.5 12.6
x1 1.5 2.5 0.5 1.2 2.6 0.3 2.4 2 0.7 1.6
x2 66 87 69 141 93 105 111 78 66 123

Correlation: The relationship between two variables such that a change in one variable results in a
 ve or -ve change in the other and also greater change in one variable results in corresponding greater
change in the other is called a correlation. For a change in one variable, if there is a corresponding change
in the other variable then the variables are called correlated.
Note:
(i) If the variables deviate in the same direction then the correlation is called direct or  ve correlation
(ii) If the variables deviate in the opposite direction then the correlation is called inverse or -ve
correlation

Correlation Coefficient (or Karl Pearson coefficient of correlation): The numerical measurement of
linear relationship between the variables x and y is called the coefficient of correlation of x and y and it
is denoted by r ( x, y) or r
Note:
(i) The coefficient of correlation r is always lies between  1 and 1 ; that is,  1  r  1
(ii) If r  0 then the variables are not correlated
(iii) If r  1 then the variables are positively and perfectly correlated
(iv) If r  1 then the variables are negatively and perfectly correlated
(v) If 0  r  1 then the variables are positively and partially correlated
(vi) If  1  r  0 then the variables are negatively and partially correlated

Correlation formulas:

Mean of x is given by x  
x
(i)
n

Variance of x is given by  x2   or  x2    (x ) 2
(x  x)2 x2
(ii)
n n

Covariance of x and y is given by Cov( x, y )   or Cov( x, y )    ( x )( y )

( x  x ) ( y  y) xy
(iii)
n n

(iv) Coefficient of correlation of x and y is given by r 

Cov( x, y )
or r 
 (x  x ) ( y  y)
 x y  ( x  x )2  ( y  y)2
 x2   y2   x2 y  x2 y   x2   y2
(v) r or r 
2 x y 2 x y
y
(vi) Regression line of y on x is given by y  y  r (x  x)
x
The slope of the regression line of y on x is called regression coefficient of y on x

It is denoted by b yx and is given by byx  r y
x

(vii) Regression line of x on y is given by x  x  r x ( y  y )
y
The slope of the regression line of x on y is called regression coefficient of x on y

It is denoted by bxy and is given by bxy  r x
y
(viii) Both the regression lines passes through the point ( x , y )
 y
(ix) The Geometric Mean of the regression coefficients is r ; that is r 2  bxy  b yx  r x  r
y x
Problems:

1. Determine the coefficient of correlation for the following data

x 1 3 4 6 8 9 11 14
y 1 2 4 4 5 7 8 9

Solution: The coefficient of correlation r 

 ( x  x) ( y  y)
 ( x  x )  ( y  y)
2 2

Here x   x  56  7 and y
 y  40  5
n 8 n 8
 x y xx y y ( x  x )( y  y ) (x  x)2 ( y  y) 2
1 1 -6 -4 24 36 16
3 2 -4 -3 12 16 9
4 4 -3 -1 3 9 1
6 4 -1 -1 1 1 1
8 5 1 0 0 1 0
9 7 2 2 4 4 4
11 8 4 3 12 16 9
14 9 7 4 28 49 16
x  y  x  x  y  y    x  x  
2
y  y 
2

56 40 84 132 56

 x  x   132, y  y   56,  x  x  y  y   84

2 2
Observe that,

Therefore, the coefficient of correlation r 

 ( x  x ) ( y  y )  84  0.977
 ( x  x)  ( y  y)
2
13256
2

2. Determine the coefficient of correlation for the following data

x 78 36 98 25 75 82 90 62 65 39
y 84 51 91 60 68 62 86 58 53 47

Solution: The coefficient of correlation r 

 (x  x ) ( y  y)
 ( x  x )  ( y  y)
2 2

Here x 
 x  650  65 and y 
 y  660  66
n 10 n 10

x y xx y y ( x  x )( y  y ) (x  x)2 ( y  y) 2
78 84 13 18 234 169 324
36 51 - 29 -15 435 841 225
98 91 33 25 825 1089 625
25 60 - 40 -6 240 1600 36
75 68 10 2 20 100 4
82 62 17 -4 -68 289 16
90 86 25 20 500 625 400
62 58 -3 -8 24 9 64
65 53 0 -13 0 0 169
39 47 - 26 -19 494 676 361
x  y  x  x  y  y    x  x  
2
y  y 
2

650 660 2704 5398 2224

 x  x   5398, y  y   2224,  x  x  y  y   2704

2 2
Observe that,

Therefore, the coefficient of correlation r 

 ( x  x ) ( y  y) 
2704
 0.7804
 ( x  x )  ( y  y)
2 2
53982224
3. Determine the coefficient of correlation for the following data

x 65 66 67 67 68 69 70 72
y 67 68 65 68 72 72 69 71

Solution: The coefficient of correlation r 

 (x  x ) ( y  y)
 ( x  x )  ( y  y)
2 2

Here x   x  544  68 and y

 y  552  69
n 8 n 8

x y xx y y ( x  x )( y  y ) (x  x)2 ( y  y) 2
65 67 -3 -2 6 9 4
66 68 -2 -1 2 4 1
67 65 -1 -4 4 1 16
67 68 -1 -1 1 1 1
68 72 0 3 0 0 9
69 72 1 3 3 1 9
70 69 2 0 0 4 0
72 71 4 2 8 16 4
x  y  x  x  y  y    x  x  
2
y  y 
2

544 552 24 36 44

 x  x   36, y  y   44,  x  x  y  y   24

2 2
Observe that,

Therefore, the coefficient of correlation r 

 ( x  x ) ( y  y) 
24
 0.603
 ( x  x )  ( y  y)
2 2
3644

4. From the following data

x 65 66 67 67 68 69 70 72
y 67 68 65 68 72 72 69 71
Determine (i) x and y (ii)  x and  y (iii) Cov( x, y) (iv) the correlation coefficient
between x and y

 x ,  2  (x  x)  ( x  x ) ( y  y)
2

Solution: We know that x x , Cov( x, y ) 

n n n
Cov( x, y )
and r 
 x y

Consider,
 x y xx y y ( x  x )( y  y ) (x  x)2 ( y  y) 2
65 67 -3 -2 6 9 4
66 68 -2 -1 2 4 1
67 65 -1 -4 4 1 16
67 68 -1 -1 1 1 1
68 72 0 3 0 0 9
69 72 1 3 3 1 9
70 69 2 0 0 4 0
72 71 4 2 8 16 4
x  y  x  x  y  y    x  x  
2
y  y 
2

544 552 24 36 44

(i) x 
 x  544  68 and y
 y  552  69
n 8 n 8

(ii)  x2 
(x  x)2 36
 4.5 ,  x  4.5  2.1213

n 8

 y2 
 ( y  y ) 2  44  5.5 ,   5.5  2.3452
y
n 8

(iii) Cov( x, y) 
 ( x  x ) ( y  y )  24  3
n 8
Cov( x, y) 3
(iv) r    0.6030
 x y 2.1213  2.3452

5. Find the correlation coefficient between x and y from the following data

x 78 89 97 69 59 79 68 57
y 125 137 156 112 107 138 123 108

Solution: The coefficient of correlation r 

 (x  x ) ( y  y)
 ( x  x )  ( y  y)
2 2

Here x   x  596  74.5 and y

 y  1006  125.75
n 8 n 8
Consider,
x y xx y y ( x  x )( y  y ) (x  x)2 ( y  y) 2
78 125 3.5 -0.75 -2.625 12.25 0.5625
89 137 14.5 11.25 163.125 210.25 126.5625
97 156 22.5 30.25 680.625 506.25 915.0625
69 112 -5.5 -13.75 75.625 30.25 189.0625
59 107 -15.5 -18.75 290.625 240.25 351.5625
79 138 4.5 12.25 55.125 20.25 150.0625
68 123 -6.5 -2.75 17.875 42.25 7.5625
57 108 -17.5 -17.75 310.625 306.25 315.0625
x  y  x  x  y  y    x  x  
2
y  y 
2

596 1006 1591 1368 2055.5

  x  x   1368, y  y  2055.5,  x  x  y  y   1591
2 2
Observe that,

Therefore, the coefficient of correlation r 

 ( x  x) ( y  y) 
1591
 0.9488
 ( x  x )  ( y  y)
2 2
13682055.5

6. From the following data

x 78 89 97 69 59 79 68 57
y 125 137 156 112 107 138 123 108
Determine
(i) x and y
(ii)  x and  y
(iii) Cov( x, y)
(iv) the correlation coefficient between x and y
(v) two regression lines

Solution: We know that x 

 x
, x 
2  (x  x)2
, Cov( x, y ) 
 ( x  x ) ( y  y)
n n n
Cov( x, y )
and r 
 x y
Consider,

x y xx y y ( x  x )( y  y ) (x  x)2 ( y  y) 2
78 125 3.5 -0.75 -2.625 12.25 0.5625
89 137 14.5 11.25 163.125 210.25 126.5625
97 156 22.5 30.25 680.625 506.25 915.0625
69 112 -5.5 -13.75 75.625 30.25 189.0625
59 107 -15.5 -18.75 290.625 240.25 351.5625
79 138 4.5 12.25 55.125 20.25 150.0625
68 123 -6.5 -2.75 17.875 42.25 7.5625
57 108 -17.5 -17.75 310.625 306.25 315.0625
x  y  x  x  y  y    x  x  
2
y  y 
2

596 1006 1591 1368 2055.5

(i) x     y  1006  125.75

x 596
 74.5 and y 
n 8 n 8

(ii)  x2 
 ( x  x )  1368  171,   171  13.0767
2

x
n 8

y 
2  ( y  y ) 2 2055.5
  256.9375 ,  y  256.9375  16.0293
n 8

(iii) Cov( x, y) 
 ( x  x ) ( y  y )  1591  198.875
n 8
Cov( x, y) 198.875
(iv) r    0.9488
 x y 13.0767 16.0293
 y
(v) The regression line of y on x is given by y  y  r (x  x)
x
16.0293
That is, y  125.75  (0.9488) ( x  74.5)
13.0767
 y  125.75  1.1630 ( x  74.5)
 y  125.75  1.1630 x  86.6435
 y  39.1065  1.163 x
x
And the regression line of x on y is given by x  x  r ( y  y)
y
13.0767
That is, x  74.5  (0.9488) ( y  125.75)
16.0293
 x  74.5  0.7740 ( y  125.75)
 x  74.5  0.774 y  97.3305
 x  22.8305  0.774 y

7. The two regression equations of the variables x and y are y  0.399 x  6.934  0 and
x  1.212 y  2.461  0. Find (i) mean of x (ii) mean of y (iii) correlation coefficient between x and y

Solution: Solving the given equations, x  11.5083 and y  11.5258

Therefore, x  11.5083 and y  11.5258
y
The regression coefficient of y on x is given by r  0.399
x

The regression coefficient of x on y is given by r x  1.212
y
    
Correlation coefficient r   r y   r x   0.3991.212  0.6954
 
 x   y 
( r is positive since both the regression coefficients are positive)

8. The two regression equations of the variables x and y are x  19.13  0.87 y and y  11.64  0.50 x.
Find (i) mean of x (ii) mean of y (iii) correlation coefficient between x and y

Solution: Solving the given equations, x  15.9349 and y  3.6726

Therefore, x  15.9349 and y  3.6726
y
The regression coefficient of y on x is given by r  0.50
x

The regression coefficient of x on y is given by r x  0.87
y
 y   x 
Correlation coefficient r   r   r    0.50 0.87   0.66
   
 x   y 

( r is negative since both the regression coefficients are negative)

  x2   y2   x2 y
9. Establish the formula r 
2 x  y
Solution: Consider,
 x2 y   x  y   x  y    x  x    y  y 
1 2 1 2

n n
  x  x     y  y    x  x  y  y 
1 2 1 2 2
n n n
  x   y  2r x y
2 2

 x2   y2   x2 y
Therefore, r 
2 x  y

 x2 y   x2   y2
10. Establish the formula r 
2 x y
Solution: Consider,
 x2 y   x  y   x  y    x  x    y  y 
1 2 1 2

n n
  x  x     y  y    x  x  y  y 
1 2 1 2 2
n n n
  x   y  2r x y
2 2

 x2 y   x2   y2
Therefore, r 
2 x y
 x2   y2   x2 y
11. Use the formula r  to compute the correlation coefficient to the following data
2 x  y

x 78 89 97 69 59 79 68 57
y 125 137 156 112 107 138 123 108

Solution: Consider,

x y xx y y (x  x)2 ( y  y) 2 z  x y zz (z  z)2

78 125 3.5 -0.8 12.25 0.64 -47 4.25 18.0625
89 137 14.5 11.2 210.25 125.44 -48 3.25 10.5625
97 156 22.5 30.2 506.25 912.04 -59 -7.75 60.0625
69 112 -5.5 -13.8 30.25 190.44 -43 8.25 68.0625
59 107 -15.5 -18.8 240.25 353.44 -48 3.25 10.5625
79 138 4.5 12.2 20.25 148.84 -59 -7.75 60.0625
68 123 -6.5 -2.8 42.25 7.84 -55 -3.75 14.0625
57 108 -17.5 -17.8 306.25 316.84 -51 0.25 0.0625
x  y  (x  x)2  ( y  y)2 z   (z  z)2
596 1006  1368  2055.52  410  241.5
 (i) x     y  1006  125.75 and z  x  y   z   410  51.25
x 596
 74.5, y 
n 8 n 8 n 8

(ii)  x2  
(x  x) 2
1368
  171 ,  x  171  13.0767
n 8

 y2 
 ( y  y ) 2  2055.52  256.94 ,   256.94  16.0293
y
n 8

 x y   z 
2 2  ( z  z ) 2 241.5
  30.1875 ,  x y   z  30.1875  5.4943
n 8
 x2   y2   x2 y 171  256.94  30.1875
(iii) r    0.9488
2 x  y 2 13.0767 16.0293
r 2  1   x y 
12. If  is the angle between the two regression lines, prove that tan 
r   x2   y2 
Solution: We know that the two regression lines
y
y y r ( x  x )   (1)
x

x  x  r x ( y  y )   (2)
y
1y
From (2), y  y  (x  x)
r x
y
The slope of the line (1), m1  r
x
1y
The slope of the line (2), m2 
r x
y 1y  r 2 1  y
  
r   x
r
m1  m2 x r x  r 2  1   x y 
Therefore, tan    
1  m1m2  y  1y
   x2   y2  r   x2   y2 
1   r  
  
 x  r x
   2 
 x 
r 2  1   x y 
If  is acute angle then tan  is positive, and therefore tan 
r  2  2 
 x y 

Exercise:

1. Determine the correlation coefficient for the following data

x 11.1 10.3 12 15.1 13.7 18.5 17.3 14.2 14.8 15.3

y 10.9 14.2 13.8 21.5 13.2 21.1 16.4 19.3 17.4 19.0

2. Compute the correlation coefficient to the following data

x 62 56 36 66 25 75 82 78
y 58 44 51 58 60 68 62 84
3. Compute the correlation coefficient to the following data

x 8 1 5 4 7
y 3 4 0 2 1

4. From the following data

x 50 60 70 90 100
y 65 51 40 26 08
Determine
(i) x and y
(ii)  x and  y
(iii) Cov( x, y)
(iv) the correlation coefficient between x and y
(v) two regression lines

5. The equations of two regression lines obtained in a correlation analysis are 4 x  5 y  33  0 and
20 x  9 y  107. Compute (i) mean of x (ii) mean of y (iii) correlation coefficient between x and y

6. Psychological tests of intelligence and engineering ability were applied to 10 students. Here is a record
of ungrouped data showing intelligence ratio (IR) and engineering ratio (ER). Calculate the coefficient
of correlation

Student A B C D E F G H I J
IR 105 104 102 101 100 99 98 96 93 92
ER 101 103 100 98 95 96 104 92 97 94

x  99, y  98,  x  x   170,   y  y   140,  x  x  y  y   92

2 2

Correlation coefficient r 
 ( x  x ) ( y  y)  0.59
 ( x  x )  ( y  y)
2 2

 x2   y2   x2 y
7. Use the formula r  to compute the correlation coefficient to the following data
2 x  y
X 62 56 36 66 25 75 82 78
Y 58 44 51 58 60 68 62 84

8. Given that x  31.6, y  38, x  3.72, y  6.31 and r  0.36 . Determine the two regression lines
Rank Correlation: The correlation between the ranks of the variables x and y is called the rank
correlation

Rank Correlation Coefficient (or Spearman Rank Correlation Coefficient): It is denoted by  ( x, y) or

6d 2
 and is given by   1  , where d is the difference between the ranks of corresponding values
n (n 2  1)
of x, y and n is the number of pairs of data points.

Repeated Values: If an item of x or y is repeated m times, then we give the average rank for the repeated
m (m 2  1)
items and add the factor to  d 2 in the formula of  .
12
Problems:
1. Determine the rank correlation coefficient for the following data

x 68 64 75 50 64 80 75 40 55 64
y 62 58 68 45 81 60 68 48 50 70

6d 2
Solution: The rank correlation of x, y is given by   1 
n (n 2  1)
The values of x in decreasing order: 80, 75, 75, 68, 64, 64, 64, 55, 50, 40
The values of y in decreasing order: 81, 70, 68, 68, 62, 60, 58, 50, 48, 45
Consider,
x y Rank x Rank y d  Rank x  Rank y d2
68 62 4 5 -1 1
64 58 6 7 -1 1
75 68 2.5 3.5 -1 1
50 45 9 10 -1 1
64 81 6 1 5 25
80 60 1 6 -5 25
75 68 2.5 3.5 -1 1
40 48 10 9 1 1
55 50 8 8 0 0
64 70 6 2 4 16
 d 2  72
Here, in the values of x, 75 is repeated 2 times and 64 is repeated 3 times
And in the values of y, 68 is repeated 2 times
Therefore the correction factor is given by
 m (m 2  1)  2(2 2  1)  3(32  1)  2(2 2  1)  1  2  1  3
12 12 12 12 2 2

 m (m 2
 1)
Now, n  10 ,  d 2  72 and d 2  12
 72  3  75
Hence the Rank correlation coefficient,
6d 2 6 (75) 450
  1  1  1  1  0.4545  0.5455
n (n  1)
2
10 (10  1)
2
990
2. Determine the rank correlation coefficient for the following data

x 10 15 12 17 13 16 24 14 22
y 30 42 45 46 33 34 40 35 39

6d 2
Solution: The rank correlation of x, y is given by   1 
n (n 2  1)
Consider,

x y Rank x Rank y d  Rank x  Rank y d2

10 30 9 9 0 0
15 42 5 3 2 4
12 45 8 2 6 36
17 46 3 1 2 4
13 33 7 8 -1 1
16 34 4 7 -3 9
24 40 1 4 -3 9
14 39 6 5 1 1
22 35 2 6 -4 16
 d 2  80
Here, there are no repetitions in the values of x and y
Now, n  9 ,  d 2  80
Hence the Rank correlation coefficient,
6d 2 6 (80) 480
  1  1  1  1  0.6667  0.3333
n (n  1)
2
9 (9  1)
2
720

3. Ten participants in a contest are ranked by two judges as follows

x 1 6 5 10 3 2 4 9 7 8
y 6 4 9 8 1 2 3 10 5 7
Calculate the rank correlation coefficient
6d 2
Solution: The rank correlation of x, y is given by   1 
n (n 2  1)
Consider,
x y Rank x Rank y d  Rank x  Rank y d2
1 6 1 6 -5 25
6 4 6 4 2 4
5 9 5 9 -4 16
10 8 10 8 2 4
3 1 3 1 2 4
2 2 2 2 0 0
4 3 4 3 1 1
9 10 9 10 -1 1
7 5 7 5 2 4
8 7 8 7 1 1
 d  60
2
 Here, there are no repetitions in the values of x and y
Now, n  10 ,  d 2  60
Hence the Rank correlation coefficient,
6d 2 6 (60) 360
  1  1  1  1  0.3636  0.6364
n (n  1)
2
10 (10  1)
2
990

4. Determine the rank correlation coefficient for the following data

x 5 10 6 3 19 5 6 12 8 2 10 19
y 8 3 2 9 12 3 17 18 22 12 17 20

6d 2
Solution: The rank correlation of x, y is given by   1 
n (n 2  1)
Consider,
x y Rank x Rank y d  Rank x  Rank y d2
5 8 9.5 9 0.5 0.25
10 3 4.5 10.5 -6 36
6 2 7.5 12 -4.5 20.25
3 9 11 8 3 9
19 12 1.5 6.5 -5 25
5 3 9.5 10.5 -1 1
6 17 7.5 4.5 3 9
12 18 3 3 0 0
8 22 6 1 5 25
2 12 12 6.5 5.5 30.25
10 17 4.5 4.5 0 0
19 20 1.5 2 -0.5 0.25
 d 2  156
Here, in the values of x, 19 is repeated 2 times, 10 is repeated 2 times, 6 is repeated 2 times, and 5 is
repeated 2 times
And in the values of y, 17 is repeated 2 times, 12 is repeated 2 times and 3 is repeated 2 times,
Therefore the correction factor is given by
 m (m 2  1)  7  2(2 2  1)  7  3.5
12 12 2

 m (m 2
 1)
Now, n  12 , d 2
 156 and d 2

12
 156  3.5  159.5
Hence the Rank correlation coefficient,
6d 2 6 (159.5) 957
  1  1  1  1  0.5577  0.4423
n (n  1)
2
12 (12  1)
2
1716
Exercise:

1. Determine the rank correlation coefficient for the following data

x 8 3 9 2 7 10 4 6 1 5
y 9 5 10 1 8 7 3 4 2 6

Ans: n  10 ,  d 2  24 and   0.8545

2. Determine the rank correlation coefficient for the following data

x 78 56 36 66 25 75 82 62
y 84 44 57 58 60 68 62 58

 m (m 2
 1)
 d 2  28.5 ,
1
Ans: n  10 , and   0.655

12 2
3. Determine the rank correlation coefficient for the following data

x 65 63 67 64 68 62 70 66 68 67 69 71
y 68 66 68 65 69 66 68 65 71 67 68 70

 m (m 2
 1)
Ans: n  12 ,  d 2  72.5 , 12
 7 and   0.722