th e s ca tte r di agr am cl early i ndi ca tes li near rel a ti onshi p

be tw een x & y . We may the re fo re w an t to fi t a st raight

l i ne to b ri ng o u t thi s rel atio ns hi p mathema ti call y

a s the poi n ts are no t on the s trai gh t li ne , the n umbe r

o f li nes tha t ca n be d raw n i s unli mi te d . T hus gi ven a
s e t o f d a ta , ou r p robl em con ce rn s ab ou t dr aw i ng the
pa r ti cul ar s trai gh t li ne th at bes t re fl ects the li nea r tren d
i ndi ca ted by the poi n ts

th e p ri nci pl e i nvol ved i n ob tai ni ng the l i ne of be st f it i s cal l ed Me thod of Le ast Square s

Supp ose y’ = a + b x i s the li ne th a t bes t fi ts the da ta . We call i t a s p redi cti on equ a ti on
E xam pl e : y1’ = a + b x 1 i s the p redi cted v al ue of y w hen x = x 1 v/s y 1 w hi ch i s the a ctual v al ue of y
w hen x = x 1
Di ffe re nce be tw een the pr edi cted v al ue & the a ctual val ue i s call ed e rro r / re si dual (e ) as i n the abo ve
ca se e 1 = y 1 – y 1 ’

(x 5 ,y 5 )
e5 y = a + bx
(x 3 ,y 3 ) ( li ne o f be s t fi t )
( x 1 ,y 1 ) e3
e1
e2
(x2,y2)

o f all the po s si bl e s tr ai gh t li nes tha t ca n be d raw n o n s ca tte r di agr am , w e ch oo se the li ne o f bes t fi t the

o ne f or w hi ch the s um o f squ are s o f e rr or s i s mi ni mum

Ba sed on tha t w e have the foll ow i ng re sul ts

- 1 -
NO TE 1 :

REG RE SSIO N EQ U ATI O NS

Y on X X on Y

y – y = byx (x – x ) x – x = bxy (y – y )

w her e : b yx & bx y a r e ca lle d as re gr e ssion coeff icien t s a nd ca n be found us ing any of the for mulae lis ted

b e low d e pe nd ing on the type of d a ta g ive n in the ques tion

b yx = ( x – x ) ( y – y ) bxy = ( x – x ) ( y – y )
2
( x – x ) ( y – y ) 2

= nx y – x y = nx y – x y

nx 2 – ( x) 2 ny 2 – ( y) 2

= nuv – u v = nuv – u v
2 2
nu – ( u) nv 2 – ( v) 2

w her e u = x – a ; v = y – b

= r y = r x

x y
NO TE 2 :

ACU TE ANG L E ( ) BE TW EE N THE TW O RE G RESSI O N LI NES

Slop e of r e gr es s ion line Y on X = m 1 = b yx

Slop e of r e gr es s ion line X on Y = m 2 = 1

bxy

ta n  = m1 – m2

1 + m1m2

NO TE 3 :

CO RRE L ATI O N COEF FI CIENT (r )

b yx . bx y = r y x r x

x y

= r2

 r =  b yx . b xy r = + ve if b yx & b xy a re p osi ti ve

r = – ve if b yx & b xy a re nega ti ve
NO TE 4 :

– 1  r  +1

 0  r2  1

 0  byx . bxy  1

- 2 -
NO TE 5 :

I f gi ven ; re gre s si on li nes Y on X a nd X an d Y

Wri te Y on X a s y = Ax + B th en A = b yx

Wri te X on Y a s x = Ay + B th en A = b xy

NO TE 6 :

I DE NTIFI C ATI O N O F REG RESS IO N LI NES

Gi ve n tw o reg res si on li ne s w i th ou t s ta ti ng w hi ch i s Y on X & w hi ch i s X on Y ; f oll ow the

s tep s to i den ti fy the reg res si on li nes

S TEP 1 : As s ume one as Y on X an d o the r a s X o n Y

S TEP 2 : U si ng N OT E 5 , fi nd by x a nd bx y

S TEP 3 : Com pu te : b yx . bx y

ST EP 4 : i f 0  b yx . bx y  1 then the as sump ti on s a re co r rect el se reve rse the as sump ti on

NO TE 7 :

Me an s : x a nd y

Si n ce the tw o re g res si on line s a re ma de to pa ss th rou gh x & y ,thei r p oi nt of i nte r secti on i s x & y

He nce to fi nd x & y ; s ol ve the tw o re g res si on l i ne s si mul ta neou sl y

NO TE 8 :

b yx + bx y  r
2

H OW : b yx = r y ; b xy = r x r x 2 + y 2  2r
x y
x .y

b yx + bx y = r y + x b yx + bx y  2r
x y

b yx + bx y  r
= r y 2 + x 2
x .y 2

FI N AL LY : PR OP OF RE GR E SSI ON COEFF.
NOW ,
a) b yx . b xy = r2
( x  y ) 2  0
b) b yx , bx y & r

x 2 + y 2 2 x .y  0 mu s t be ar the same si gn

b) 0  b yx .bx y  1
x 2 + y 2  2 x .y
c) i f by x  1 then b xy  1
x 2 + y 2  2
d) b yx + bx y  r
x .y 2

- 3 -
NO TE 8 :

Change of scale affects the regressi on coefficients then ‘How to get back to ori ginal

regression coefficients’

Let u = x – a & v = y – b
c d

You compute the regressi on coefficients usi ng

b vu = n uv – u v & bu v = n uv – u v

n u 2 – ( u ) 2 n v 2 – ( v ) 2

To return to ori ginal regressi on coefficients

byx = d bvu & bxy = c buv

c d

- 4 -
1. Obtai n the regression lines
2. Obtai n the regression line y on x
x y x – x y – y (x – x ) 2 (y – y ) 2 (x – x )(y – y )
x y x – x y – y (x – x ) 2 (y – y ) 2 (x – x )(y – y )
1 5  2  4 4 16 8
3 10  4  8 16 32
2 7  1  2 1 4 2
5 14  2  4 4 8
3 9 0 0 0 0 0
7 18 0 0 0 0
4 11 1 2 1 4 2
9 22 2 4 4 8
5 13 2 4 4 16 8
11 26 4 8 16 32
0 0 10 40 20
0 0 40 80
x = 3 y = 9  (x – x ) 2  (y – y ) 2  (x – x )(y – y )
x = 7 y = 18  (x – x ) 2  (y – y ) 2  (x – x )(y – y )

Y ON X X ON Y
Y 0N X
byx =  (x – x )(y – y ) bxy =  (x – x )(y – y )
byx =  (x – x )(y – y )
 (x – x ) 2  (y – y ) 2
 (x – x ) 2
= 20 = 20
10 40 = 80
40
= 2 = 0.5
= 2

y – y = byx (x – x ) x – x = bxy (y – y )
y – y = byx (x – x )

y – 9 = 2 (x – 3 ) x – 3 = 0.5(y – 9 )
y – 18 = 2 (x – 7)

y – 9 = 2x – 6 x – 3 = 0.5x – 4.5
y – 18 = 2x – 14
y = 2x – 6 + 9 x = 0.5x – 4.5 + 3
y = 2x – 14 + 18

y = 2x + 3 x = 0.5x – 1.5
y = 2x + 4

- 5 -
3. x : years of servi ce
y : income (000’s) 04
y x
X : 21 25 26 24 19
a) Find regression equation of income on the years of service
b) What i niti al start would you recommend for a person Y : 19 20 24 21 16
applying for a job after having served in si mi lar capaci ty in Obtai n regressi on line X on Y
another company for 13 years

x y x – x y – y (x – x ) 2 (y – y ) 2 (x – x )(y – y )
SOLUTION
11 10 3 2 9 6

7 8  1 0 1 0 x y x – x y  y (y – y ) 2 (xx )(yy )

9 6 1  2 1  2 21 19 2 1 1 2

5 5  3  3 9 9 25 20 2 0 0 0

26 24 3 4 16 12
8 9 0 1 0 0
24 21 1 1 1 1
6 7 2  1 4 2
19 16 4 4 16 16
10 11 2 3 4 6
x = 23 0 34 31
56 56 0 0 28 21
y = 20
x y
 (x – x ) 2  (y – y ) 2  (x – x )(y – y )
x = 7 y = 7
bxy =  (x – x )(y  y )
 (y – y ) 2
byx =  (x – x )(y – y ) y – y = byx (x – x )
 (x – x ) 2 = 31
y  8 = 0.75(x – 8) 34
= 21
28 y – 8 = 0.75x – 6
= 0.91 (DI RECT DI VI SI ON)
= 3 y = 0.75x – 6 + 8
4 x – x = bxy (y – y )
y = 0.75x + 2 x  23 = 0.91(y – 20)
= 0.75 Put x = 13
x – 23 = 0.91y  18.2
y = 0.75(13) + 2
= 9.75 + 2 x = 0.91y + 4.8

= 11 .75(000 ’ s)

- 6 -
5. x : index of producti on 6. x : test scores
y : no. of unemployed (i n lacs) y : sales (i n 00’s)

find regression line y on x a) find regression line y on x b) Esti mate y when x = 17

x y x – x y – y (x – x ) 2 (y – y ) 2 (x – x )(y – y ) x y x – x y – y (x – x ) 2 (y – y ) 2 (x – x )(y – y )

100 15  4 0 16 0 16 35  7  10 49 70

102 12  2  3 4 6 22 42  1  3 1 3

104 13 0  2 0  0 28 57 5  12 25 60

107 11 3  4 9 12 24 40 1  5 1 5

105 12 1  3 1  3 29 54 6  9 36 54

112 12 8  3 64 24 25 51 2  6 4 12

103 19  1 4 1  4 16 34  7  11 49 77

99 26  5 11 25 55 23 47 0 2 0 0

832 120 0 0 120  98 + 6 =92 24 45 1 0 1 0

x y
 (x – x ) 2  (y – y ) 2  (x – x )(y – y ) 207 405 0 0 166 276 – 5 = 271
x =10 4 y = 15
x y  (x – x ) 2  (y – y ) 2  (x – x )(y – y )

x =23 y = 45

byx =  (x – x )(y – y ) y – y = byx (x – x )

b yx = ( x – x ) (y – y ) y – y = b yx (x – x )
 (x – x ) 2 2
( x – x )
y  15 = 0.77 (x – 104)
y  45 = 1 .632 ( x – 23)
= 92
= 271
120 y  15 = 0.77x + 80.08
166 y  45 = 1 .632 x  37 .536

= 0.77 y = 0.77x + 80.08 + 15

LOG CALC = 1 .632 y = 1 .632 x + 7 .464

( DI RE CT DI VI SI ON ) y = 0.77x + 95.08
2 .4330 y = 1 .632(17 ) + 7 .464
 2 .2201
A L 0 .2129 y = 27 .744 + 7 .464
1 .632
y = 35 .208

- 7 -
07. Y 0N X X 0N Y
The f oll ow ing ta bl e gi ve s th e ap ti tude s co res and p ro du cti vi ty i ndi ces of 10
byx =  (x – x )(y – y ) bxy =  (x – x )(y – y )
w orke r s sel ecte d a t ra ndom
 (x – x ) 2  (y – y ) 2
Ap ti tude : 60 62 65 70 72 48 53 73 65 82
Sco re (X ) = 1044 = 1044
894 1752
P ro du cti vi ty : 68 60 62 80 85 40 52 62 60 81
I ndex (Y )
LOG CALC LOG CALC
a) O b tai n the tw o reg re ssi on li nes
3 .0187 3 .0187
b) E s ti ma te the p rodu cti vi ty in dex of a w orke r w i th te s t sco re 9 2  2 .9513  3 .2435
c) th e te s t s core w hose p rodu cti vi ty i ndex i s 7 5 A L 0 .0674 A L 1 .7752
1 .168 0 .5960

x y x – x y – y (x – x ) 2 (y – y ) 2 (x – x )(y – y )
= 1.168 = 0.5960
60 68  5 3 25 9 15
y – y = byx (x – x ) x – x = bxy (y – y )
62 60  3  5 9 25 15

65 62 0  3 0 9 0
y – 65 = 1.168 ( x – 65) x – 65 = 0.596 ( y – 65)

70 80 5 15 25 225 75 y  65 = 1.168x  75.92 x  65= 0.596 y  38.74

72 85 7 20 49 400 140
y = 1.168x  75.92 + 65 x = 0.596y  38.74 + 65
48 40 17 25 289 625 425
y = 1.168 x  10.92 x = 0.596 y + 26.26
53 52 12 13 144 169 156
Put x = 92 Put y = 75
73 62 8  3 64 9  24

65 60 0  5 0 25 0 y = 1.168 (92)  10.92 x = 0.596 (75) + 26.26

82 81 17 16 289 256 272 y = 107.456  10.92 x = 44.7 + 26.26

650 650 0 0 894 17 52 1083–39 =1044
y = 96.536 x = 70.96
x y  (x – x ) 2  (y – y ) 2  (x – x )(y – y )
Producti vity index = 96.536 test score = 70.96
x =65 y = 65
When test score = 92 when producti vi ty index = 75

- 8 -
08. Two samples from bi vari ate populati ons have 15 observati ons
09. from the data of 20 pairs of observati ons on X and Y following
each . The sample means of X and Y are 25 and 18
results are obtained
respecti vely . The corresponding sum o f squares of devi ations
from means are 136 and 148 . The sum of product of devi ations x = 199 ; y = 94 ; (x – x ) 2 = 1298 ;  (y – y ) 2 = 600
from respecti ve means i s 122 . Obtain the equati on of line of
 (x – x )(y – y ) = – 262 .
regression of X on Y
find
SOLUTI ON a) the line of regressi on of y on x b)line of regressi on of x on y
x = 25 , y = 18 ,  (x – x ) 2 = 136 ;  (y – y ) 2 = 148
SOLUTION

 (x – x )(y – y ) = 122
Y 0N X X 0N Y

X 0N Y byx =  (x – x )(y – y ) bxy =  (x – x )(y – y )

 (x – x ) 2  (y – y ) 2
bxy =  (x – x )(y – y )
LOG CALC = 262 = 262
 (y – y ) 2
2 .0864 1298 600
= 122  2 .1703
148 A L 1 .9161 LOG CALC LOG CALC
0 .8243
2 .4183 2 .4183
= 0.8243  3 .1134  2 .7782
A L 1 .3049 A L 1 .6401
0 .2018 0 .4366
x – x = bxy (y – y )

= 0.2018 = 0.4366
x – 25 = 0.8243 (y – 18)
y – y = byx (x – x ) x – x = bxy (y – y )
x – 25 = 0.8243 y – 14.8374
y – 94 = 0.2018 ( x – 199 ) x – 199 = 0.4366 ( y – 94)

x = 0.8243 y – 14.8374 + 25 y  94 = 0.2018x + 40.1582 x  199 = 0.4366y + 41.0404

y = 0.2018 x + 40.1582 + 94 x = 0.4366y + 41.0404 + 199

x = 0.8243 y + 10.1626
y = 0.2018 x + 134.1582 x = 0.4366y + 240.040 4

- 9 -
10. for a bi variate data ; STEP 1

x = 53 , y = 28 , byx =  1.5 , bxy = 0.2 byx =  (x – x )(y – y ) = 6 = 1

Find a) esti mate of y for x = 50  (x – x ) 2 30 5

b) esti mate of x for y = 25

SOLUTION bxy =  (x – x )(y – y ) = 6 = 1

 (y – y ) 2 36 6
a) Y ON X

y  y = byx ( x – x ) STEP 2

y – 28 = 1.5(x – 53)
m 1 = byx = 1 ; m2 = 1 = 6
y  28 = 1.5(50 – 53) 5 bxy

y  28 = 1.5(3)

y  28 = 4.5 STEP 3

y = 32.5 tan  = m1 – m2
1 + m 1 .m 2

b) X ON Y 1  6
tan  = 5 2
x  x = bxy ( y – y ) 1 + 1 6
5 2
x – 53 = 0.2(x – 28)
tan  = 1 – 30
x  53 = 0.2(25 – 28)
5 + 6
x  53 = 0.2(3)
tan  = 29
x  53 = 0.6 11

x = 53.6 tan  = 29
11

09.  x = 40 ;  y = 32 ;  (x – x ) 2 = 30 ,  (y – y ) 2 = 36 ;  = tan  1 29
11
 (x – x )(y – y ) = 6 , find the regression coefficient s and the

measure of the acute angle between the regressi on line

- 10 -
 01 . find the reg ress ion line y on x . Hence find y
w hen x = 10

x y x2 xy

1 2 1 2

2 4 4 8

3 7 9 21

4 6 16 24

5 5 25 25

6 6 36 36

21 30 91 116
2
x y x x y

x = x = 21 = 3.5
n 6

y = y = 30 = 5
n 6

byx = n xy   x. y
n x 2  ( x) 2

= 6(116)  (21)(30)
6 (91)  (21) 2

= 696  630
546  441

= 66
105

= 0.63

Equation

y  y = byx (x – x )

y  5 = 0.63( x  3.5)

y  5 = 0.63 x  2.205

y = 0.63 x  2.205 + 5

y = 0.63 x + 2.795

Put x = 10

y = 0.63(10) + 2.795

y = 6.3 + 2.795 = 9.095

11
02 . H e ig ht of fa ther (X ) : 64 62 66 63 67 61 69 65 Y on X X on Y

H e ig ht of Son (Y) : 67 65 67 64 68 65 67 64
b yx = b vu bxy = b uv
H e ig ht of fa ther (X ) : 67 66

H e ig ht of Son (Y) : 70 66 = nuv  uv = n uv  uv

a ) F ind the r eg r es sion line s nu 2  ( u) 2 n v 2  ( v) 2

b ) Es tima te s ons heig ht if fa the r s he ig ht is 68 inc hes = 10(27 )  0(3) = 10(27 )  0(3)
2
c ) Es tima te fa the rs he ig ht if s ons he ig ht is 59 inc h es 10(56 ) – (0) 10(33 ) – (3) 2

= 27 = 27
X y u= x65 v= y66 u2 v2 uv 56 321

64 67 1 1 1 1 1 = 0 .4821 = 0 .8412

62 65 3  1 9 1 3
R eg r es s ion line Y on X R eg r es s ion Line X on Y
66 67 1 1 1 1 1
y – y = b yx( x – x) x – x = b x y( y – y)
63 64 2  2 4 4 4
y – 66 .3 = 0 .4821( x – 65) x – 65 = 0 .8412( y – 66 .3 )
67 68 2 2 4 4 4

61 65 4  1 16 1 4 y – 66 .3 = 0 .4821 x – 31 .3365 x – 65 = 0 .8412 y – 55 .77156

69 67 4 1 16 1 4 y = 0 .4821 x  31 .3365 + 66 .3 x = 0 .8412 y  55 .771 56 + 65

65 64 0  2 0 4 0
y = 0 .4821 x + 34 .9635 x = 0 .8412 y + 9 .22844
67 70 2 4 4 16 8

66 66 1 0 1 0 0 P ut x = 68 P ut y = 59

650 663 0 3 56 33 27
y = 0 .4821(68 ) + 34 .9635 x = 0 .8412 (59 ) + 9 .22844
x y u v u 2 v 2 uv
y = 32 .7828 + 3 4.9 635 x = 49 .6308 + 9 .22 844

x = x = 650 = 65 y = 67 .7463 x = 58 .8592

n 10

y = y = 663 = 66 .3 Sons heig ht = 67 .7463 inc hes fa ther ’s height = 58 .8592 inc hes
n 10
When fa thers height = 68 inc hes w hen son s heig ht = 59 inc hes

- 12 -
03.
04.
x i = 56 ; y i = 56 ; x i 2 = 476 ; y i 2 = 476 ;
I nformati on on vehicles (i n thousands)
x i y i = 469 , n = 7
passing through seven different highways
a) Obtai n linear regression of Y on X
during a day (X) and number of acci dents
b) y if x = 12
reported (Y) i s gi ven as

x = x = 56 = 8 x = 105 ; y = 409 ; x 2 = 1681 ;y 2 = 39350

n 7
xy = 8075
y = y = 56 = 8
n 7 Obtai n linear regression of Y on X

byx = n xy   x. y SOLUTION

n x 2  ( x) 2
x = x = 105 = 15
n 7
= 7(469)  (56)(56)
7(476)  (56) 2 y = y = 409 = 58.43
n 7
= 3283  3136
3332  3136 byx = n xy   x. y

= 147 n x 2  ( x) 2

196
= 7(8075)  (105)(409)
= 21
7 (1681)  (105) 2
28
= 56525  42945
= 3 11767  11025 LOG CALC
4
4 .1329
= 13580  2 .8704
= 0.75
742 A L 1 .2625
Equation
18 .30
= 18.30
y  y = byx (x – x )

Equation
y  8 = 0.75( x  8)

y  y = byx (x – x )
y  8 = 0.75 x  6

y = 0.75 x + 2 y  58.43 = 18.30 ( x  15)

Put x = 12
y  58.43 = 18.30x  274.5

y = 0.75(12) + 2
y = 18.30x  274.50 + 58.43
y = 9 + 2

y = 11 y = 18.30x  216.07

13
05. 06.
n = 10 ,  x = 250 ,  y = 300 ,  x 2 = 6500 ,
The records of ten days is gi ven
 y 2 =10000 ,  xy = 7900 .
x = 580 ; y = 370 ; x 2 = 41658 ; Find acute angle between two lines
STEP 1
y 2 = 17206 ; xy = 11494

Obtai n regressi on line x on y byx = n xy   x. y

n x 2  ( x) 2
SOLUTION

= 10 (7900 )  250(300)
bxy = n xy   x. y 10(6 500)  (250) 2
n y 2  ( y) 2
= 79000  75000
= 10(11494)  (580)(370) 65000 – 62500

10 (17206)  (370) 2
= 4000
2500
= 114940  214600
172060  136900 = 85
LOG CA LC

= 99660 3 .9986 bxy = n xy   x. y

35160  3 .5460
n y 2  ( y) 2
A L 0 .4526
=  9966 2 .835
3516 = 10 (7900)  250(300)
10(10000)  (300) 2
=  2.835
= 79000  75000
100000 – 90000
x  x = bxy ( y – y )
= 4000
x – 58 = 2.835 (y – 37) 10000

x  58 = 2.835y + 104.895 = 25

STEP 2
x = 2.835y + 162.895
m 1 = byx = 8 , m2 = 1 = 5
5 bxy 2

STEP 3

tan  = m1 – m2
1 + m 1 .m 2

8  5
tan  = 5 2
1 + 8 5
5 2

tan  = 16 – 25
10 + 40

tan  = 9
50

 = tan  1 9
50

14
01. you are gi ven below the following information about adverti sing
02. Sales (crores) exp.(crores)
and sales (x) (y )
Ad v. Exp. (x) sales (y ) in lacs Mean 40 6
Mean 10 90 S.D 10 1.5 correlati on coefficient = 0.9
S.D. 3 12 r = 0.8 . Obtai n regressi on lines to
a) What is the likely sales when adv. budget is ` 15 lacs a) esti mate sales for a proposed adv. exp. of ` 10 crores

b) what should be adv. exp. if the company wants sales target b) esti mate adv. exp. for proposed sales target of ` 60 crore
y on x x on y
of ` 120 lacs
y on x x on y byx = r . y bxy = r . x
x y
byx = r . y bxy = r . x
x y = 0.9 x 1.5 = 0.9 x 10
10 1.5
= 0.8 x 12 = 0.8 x 3
3 12 = 1.35 = 9 x 10
10 15
= 0.8 x 4 = 3.2 = 0.8 = 0.2
4 = 0.135 = 6

y – y = byx(x – x ) x – x = bxy(y – y ) y – y = byx(x – x ) x – x = bxy(y – y )

y – 90 = 3.2 (x – 10) x – 10 = 0.2(y – 90) y – 6 = 0.135(x – 40) x – 40 = 6 (y – 6)

y – 90 = 3.2x – 32 x – 10 = 0.2y – 18 y – 6 = 0.135x – 5.4 x – 40 = 6y – 36

y = 3.2x – 32 + 90 x = 0.2y – 18 + 10 y = 0.135x – 5.4 + 6 x = 6y – 36 + 40

y = 3.2x + 58 x = 0 .2y  8 y = 0.135x + 0.6 x = 6y + 4

Put x = 15 Put y = 120 Put x = 60 Put y = 10

y = 3.2(15) + 58 x = 0.2(120)  8 y = 0.135 (60) + 0.6 x = 6(10) + 4

y = 48 + 58 x = 24 – 8 y = 8.1 + 0.6 = 8.7 x = 60 + 4 = 64

y = 106 x = 16
adv exp. = ` 8.7 cr when sales = ` 64 cr when
Sales = ` 106 lacs when adv exp. = ` 16 lacs when Sales = ` 60 cr adv exp. = ` 10 cr
Ad v. budget = ` 15 lacs sales target = ` 120 lacs

- 15 -
03. you are gi ven below the following information about Profi t rate

(X) and growth rate (Y) 04. if for bi – variate data x = 10 and y = 12 , vari ance V(x ) = 9 ;
X Y
y = 4 & r = 0.6
Mean 20 25
Variance 9 25 correlati on coeff. = 0.9 . Esti mate x when y = 10 and y when x = 5
y on x x on y
a) Obtai n the regression line of X on Y
b) Esti mate the growth rate when the profi t rate i s 16% byx = r . y bxy = r . x
y on x x on y x y

byx = r . y bxy = r . x = 0.6 x 4 = 0.6 x 3

x y 3 4

= 0.9 x 5 = 0.9 x 3 = 0.2 x 4 = 0.3 x 3

3 5 2

= 0.3 x 5 = 2.7 x 2 = 0.8 = 0.45

5 x 2

= 1.5 = 0.54 y – y = byx(x – x ) x – x = bxy(y – y )

y – y = byx(x – x ) x – x = bxy(y – y ) y – 12 = 0.8 (x – 10) x – 10 = 0.45 (y – 12)

y – 25 = 1.5 (x – 20) x – 20 = 0.54 (y – 25) y – 12 = 0.8x – 8 x – 10 = 0.45 y – 5.4

y – 25 = 1.5x – 30 x – 20 = 0.54 y – 13.5 y = 0.8x – 8 + 12 x = 0.45 y – 5.4 + 10

y = 1.5x – 30 + 25 x = 0.54 y – 13.5 + 20 y = 0.8x + 4 x = 0.45 y + 4.6

y = 1.5x  5 x = 0.54 y + 6.5 Put x = 5 Put y = 10

Put x = 16 y = 0.8(5) + 4 x = 0.45(10) + 4.6

y = 1.5(16)  5 y = 4 + 4 x = 4.5 + 4.6

y = 24  5 = 19 y = 8 x = 9.1

growth rate = 19%

when the profi t rate i s 16%

16
05. Y on X
from the following data regarding the
byx = r y
wholesale and retail price of a commodi ty
x
esti mate the retail price when wholesale
price is Rs 240 per quintal = 0.93 x 2
3
Wholesale(x) Retail (y)
= 0.62
Avg price 200 280
(per qtl.) y  y = byx (x – x )
S.D. 20 25
y – 23 = 0.62(x – 27)
r = 0.94
Put x = 23
Y on X
y – 23 = 0.62(23 – 27)
byx = r y
x y – 23 = 0.62(4)

= 0.94 x 25 y – 23 = 2.48
20
y = 20.52

= 1.175
ans : age of wife = 20.52 when age of
y  y = byx (x – x )
husband = 23 yrs

y – 280 = 1.175 (x – 200)

Put x = 240
07. Find the equation of line of regression of
y – 280 = 1.175 (240 – 200)
Y on X for the following data
y – 280 = 1.175 (40)
n = 8 ;  (x – x )(y – y ) = 120 ; x = 20
y – 280 = 47
y = 36 ; x = 2 ; y = 3

y = 327 byx = cov(x,y)

x 2
ans : Retail price = Rs327/ - when whole sale
 ((x – x )(y – y )
price is Rs 240-
= n
x 2

06. 120
= 8 = 15 = 3.75
the following data relates to the age of
4 4
husband and wife Y on X

Husbands Wi ves y – y = byx (x – x )

X Y y – 36 = 3.75(x – 20)

Mean 27 yrs 23 yrs y – 36 = 3.75x – 75

SD 03 yrs 02 yrs ; r = 0.93 y = 3.75x – 75 + 36

Esti mate the age of the woman whose
y = 3.75x – 39
husband i s aged 23

- 17 -
08. The followi ng results were obtained from

records of age (X) and systolic blood

pressure (Y) of 10 women

X Y

Mean 53 142

Vari ance 130 165 ;

 (x – x )(y – y ) = 1220

Find regression equati on to esti mate 09

the blood pressure of a woman w i th the following results were obtained from

age 47 years record of age X and systolic blood pressure (Y)

of a group of 10 men
byx = cov(x,y) X y
x 2
Mean 50 140

 ((x – x )(y – y ) Vari ance 130 165 ,  (xx )(yy ) = 1220

= n Obtai n the regressi on line to esti mate blood
x 2 pressure of a man of age 40 years

SOLUTION
1220 LOG CA LC
= 10
130 2 .0864 byx = cov(x,y)
 2 .1139
x 2
= 122 A L 1 .9725
130 0 .9387  ((x – x )(y – y )
= n
= 0.9387 x 2
Y on X
1220
= 10
y – y = byx (x – x ) LOG C ALC
130
y – 142 = 0.9387(x – 53) 2.0864
= 122  2.1139
y – 142 = 0.9387x – 48.7511 130 AL 1.9725
0.9387
y = 0.9387x – 48.7511 + 142 = 0.9387

y = 0.9387x + 92.2489 Y on X

y – y = byx (x – x )
Put x = 47
y – 140 = 0.9387(x – 50)
y = 0.9387(47) + 92.2489
y – 140 = 0.9387x – 46.935
y = 44.1189 + 92.2489
y = 0.9387x + 93.065
= 136.3678
Put x = 40
ans : esti mated blood pressure of a
y = 0.9387(40) + 93.065
woman aged 47 years is 136.3678
y = 37.548 + 93.065

= 130.613

ans : esti mated blood pressure of a

man aged 40 years i s 130.613

- 18 -
10. I f for a bi variate data byx = 1.2 ,
01. The equation of the two regression lines
x = 4 ; r = 0.6 find y , bxy and the
are 3x + 2y – 26 = 0 and 6x + y – 31 = 0 .
acute angle between the two regressi on
a) correlati on coefficient b)mean’s
li nes
ST EP 1
ST E P 1 b yx = r . y
x
AS SUME
1.2 = 0.6 x y
4 YON X : 3x + 2y – 26 = 0

y = 1.2 x 4 2y =  3x + 26
0.6 y = 8
y = 3x + 26
STEP 2
bxy = r . x 2 3
y
byx =  3
= 0.6 x 4 = 0.3 2
8
STEP 3 X ON Y : 6x + y – 31 = 0

m 1 = byx = 1.2 = 12 = 6 6x =  y + 31

10 5
x =  1 y + 31
6 6
m2 = 1 = 1 = 10
bxy 0.3 3 bxy = 1
6
ST E P 4
STEP 2
tan  = m1 – m2
1 + m 1 .m 2 r2 = byx . bxy

6  10 = 3 x 1
tan  = 5 3 2 6

1 + 6 10
= 1
5 3
4

tan  = 18 – 50
Since 0  r2  1
15 + 60
Our assumpti ons are correct

tan  = 32
r =  1
75
2

 = tan  1 32 =  1 ( byx & bxy are ve)

57 2

STEP 3

2 x 3x + 2y = 26
6x + y = 31

6x + 4y = 52
 6x + y = 31
3y = 21

y = 7

subs in (1) x = 4

- 19 -
 03. the equati ons of the two regressi on lines
02. the regression equations are
are 10x – 4y = 80 & 10y – 9x =  40
10x + 3y – 62 = 0 & 6x + 5y –50 = 0
Find i ) means ii ) byx & bxy
I dentify the regressi on of x on y . Hence
iii) var (x) if var(y) = 36 i v) r
find mean of x & y . Als o fi nd x if y = 2 STEP 1

STEP 1 AS SUME

AS SUME XON Y : 10x  4y = 80

XON Y : 10x + 3y – 62 = 0 10x = 4y + 80

10x=  3y + 62 x = 4y + 80
10 10
x = 3y + 62
10 10 bxy = 4
10
bxy =  3
10 Y ON X: 10y – 9x =  40

Y ON X: 6x + 5y – 50 = 0 10y = 9x  40

5y =  6x + 50 y = 9 x  40
10 10
y =  6 x + 50
5 6
byx = 9

byx = 6 10
5
STEP 2
STEP 2

r2 = byx . bxy
r2 = byx . bxy
= 4 x 9
= 3 x 6
10 5 10 5

= 9 = 36
25 100

Since 0  r2  1 Since 0  r2  1
Our assumpti ons are correct Our assumpti ons are correct

r =  3 r =  6
5 10

=  3 ( byx & bxy are ve) = + 3 ( byx & bxy are + ve)
5 5

STEP 3 STEP 4 STEP 3 STEP 4

10x + 3y = 62 bxy = r x x 10x  4y = 80 x 5 bxy = r x x

6x + 5y = 50 y 9x + 10y = 40 x 2 y

50x + 15y = 310  3 = 3 x x 50x  20y = 400 4 = 3 x x

 18x + 15y = 150 10 5 2  18x + 20y = 80 10 5 6
32x = 160 32x = 320
x = 1
x = 5 x = 10 x = 4

subs in (2) y = 4 subs in (2) y = 5 var (x) = 16

- 20 -
04. vari ance of x = 9 , 05. For the two regressi on equati ons :

Reg. equ ’s 8x – 10y + 66 = 0 ; 40x – 18y = 214 4y = 9x + 15 ; 25x = 6y + 7

Find the mean of x and y and standard find correlati on coefficient r ; x and y
STEP 1
devi ati on of y
STEP 1
AS SUME

AS SUME
YON X : 4y = 9x + 15

XON Y : 40x – 18y = 214 y = 9x + 15

40x = 18y + 214 4 4

x = 18y + 214 byx = 9

40 40 4

X ON Y : 25x = 6y + 7
bxy = 18 = 9
40 20 x = 6 y + 7
25 25
Y ON X: 8x – 10y + 66 = 0
bxy = 6
10y = 8x + 66
25
y = 8 x + 66
STEP 2
10 10

r2 = byx . bxy
byx = 8 = 4
10 5 = 9 x 6

STEP 2 4 25

= 54
r2 = bxy . byx
100
= 9 x 4
20 5 Since 0  r2  1
Our assumpti ons are correct
= 36
r =  54
100
100
0  r2  1 Since
r = + 54 ( byx & bxy are ve)
Our assumpti ons are correct
100
r =  6
10 log r = 1 log 54 – log 100
2
= + 3 ( byx & bxy are + ve)
5 log r = 1 1.7324  2.0000
STEP 3 STEP 4 2

40x – 18y = 214 byx = r x y log r = 1.7324  2.0000

8x – 10y = 66 x 5 x 2 2

40x – 18y = 214 4 = 3 x y log r = 0.8662  1.0000

40x  50y = 330 5 5 3
 + +
log r = 1 . 8662
32y = 544 y = 4

y = 17 r = AL (1 . 8662 )

subs in (2) x = 13 var (y) = 16 r = 0.7348

- 21 -
 06. Regression of two series are

2x – y – 15 = 0 & 3x – 4y + 25 = 0

F ind mea n of x & y ,c oefficient of c or rela tion

STEP 1

AS SUME

XON Y : 2x – y – 15 = 0

2x = y + 15

x = 1 y + 15
STEP 3
2 2

3 x 9x – 4y =  15
bxy = 1
2 x 25x  6y = 7 2

27x  12y = 45 Y ON X : 3x – 4y + 25 = 0

50x  12y = 14 4y = 3x + 25
 + 
 23x = 59 y = 3 x + 25
4 4
x = 59
23 byx = 3
4
subs in (1) STEP 2

9 59  4y =  15 r2 = bxy . byx
23
= 1 x 3
531  4y =  15 2 4
23
= 3
 4y =  15  531 8
23
Since 0  r2  1
 4y =  345  531 Our assumpti ons are correct
23
r =  3
 4y =  876 8
23
r = + 3 ( byx & bxy are + ve)
4y = 876 8
23
log r = 1 log 3 – log 8
2
y = 219
23 log r = 1 0.4771  0.9031
2

log r = 0.477 1  0.9031

2 2

log r = 0.2386  0.4516

log r = 1 . 7870

r = AL (1 . 7870 )

r = 0.6124

- 22 -
07. 2x + 3y – 6 = 0 & 5x + 7y – 12 = 0 are S TEP 3

two regre ssi on lines .Fi nd correlation coefficient

bxy = r. x
STEP 1 y

AS SUME  7 = 0.9661 x

XON Y : 5x + 7y – 12 = 0 5 y

5x =  7y + 12 x = 7
y 5 x 0.9661
x = 7y + 12
5 5
x = 7
bxy =  7
y 4.8305 LOG CALC
5 4.8451
 4.6840
Y ON X : 2x + 3y – 6 = 0 x = 70000
y 48305 AL 0.1611
3y =  2x + 6 1.449
x = 1.449
y =  2 y + 6
3 3 y

byx = 2
3
STEP 2 08. Values of two regressi on coefficients

between the vari ables X and Y are byx = –

r2 = bxy . byx
0.4 and bxy = – 2.025 respecti vely . Obtain
= 7 x 2
the value of correlati on coefficient
5 3
SOLUTION
= 14
15 r 2 = byx x bxy

Since 0  r2  1
r 2 = 0.4 x 2.025
Our assumpti ons are correct
r2 = 4 x 2025
r =  14
10 1000
15

r =  14 ( byx & bxy are ve) r2 = 8100

10000
15

r2 = 81
log r ’ = 1 log 14 – log 15 100
2
r =  9
log r’ = 1 1.1461  1.1761 10
2
r =  9 (byx & bxy are ve)
10
log r’ = 1.1461  1.1761
2 2

log r’ = 0.5730  0.5880

log r ’ = 1 . 9850

r ’ = AL (1 . 9850 )

r’ = 0.9661 r =  0.9661

- 23 -
09. Values of two regression coefficients
(byx and bxy) are –0 .75 and –1.1
respecti vely . Obtain the value of r

SOLUTION

r 2 = byx x bxy

r 2 = 0.75 x 1.1

10. regression coefficients between the

r2 = 75 x 11
100 10 vari ables X and Y are byx = –0.6 and bxy = –

0.3 respecti vely . Obtain the value of

r2 = 825
correlati on coefficient
1000
SOLUTION

r =  825
1000 r 2 = byx x bxy

r 2 = 0.6 x 0.3
r =  825 ( byx & bxy are ve)
1000
r2 = 6 x 3
10 10
log r’ = 1 log 825 – log 1000
2 r2 = 18
100
log r’ = 1 2.9165  3.0000
2
r =  18
100
log r’ = 2.9165  3.0000
2 2
r =  18 ( byx & bxy are ve)
log r’ = 1.4583  1.5000 100

log r ’ = 1 . 9583 log r’ = 1 log 18 – log 100

2
r ’ = AL (1 . 9583 )
log r’ = 1 1.2553  2.0000
2
r’ = 0.9084

log r’ = 1.2553  2.0000

r =  0.9084
2 2

log r’ = 0.6277  1.0000

log r’ = 1 . 6277

r’ = AL (1 . 6277 )

r’ = 0.4243

r =  0.4243

- 24 -
11. regression coefficients between the 12. for 50 students of a class the regression

vari ables X and Y are byx = – 1.5 and bxy equati on of marks in Stati stics (x) on

= – 0.2 respecti vely . Obtai n the value the marks in a/c (y) i s 3y –5x+180 = 0 .

of correlation coefficient The mean of marks of accounts is 44

SOLUTION and vari ance of marks in Stati stics is

9/16 t h of the vari ance of marks in

r 2 = byx x bxy
accounts . Find mean marks of
r 2 = 1.5 x 0.2 Stati stics and correlati on coefficient

r2 = 15 x 2
SOLUTION :
10 10
GIVEN : X ON Y : 3y – 5x + 180 = 0
r2 = 30 y = 44
100
x 2 = 9

r =  3 ( byx & bxy are ve) y 2 16

10 STEP 1

X ON Y : 3y – 5x + 180 = 0
log r’ = 1 log 3 – log 10
5x = 3y + 180
2
x = 3y + 180
log r’ = 1 0.4771  1.0000 5 5
2
bxy = 3
5
log r’ = 0.4771  1.0000
2 2 STEP 2
bxy = r . x
log r’ = 0.2386  0.5000 y

3 = r x 3
log r’ = 1 . 7386
5 4

r’ = AL (1 . 7386 ) r = 4
5

r’ = 0.5478 STEP 3
Put y = 44 i n
r =  0.5478
X = 3y + 180
5 5

x = 3 (44) + 180
5

x = 132 + 180
5

x = 312
5

x = 62.4

mean marks in stati stics = 62.4

- 25 -
13. For group of 100 students the equation
4  25
of regressi on line of marks i n Economics tan  = 3 12
(Y) on marks in Book keeping (X) is gi ven 1 + 4 25
3 12
by 4x – 3y – 120 = 0 . Rati o of vari ance

of x to variance of y is 9 : 25 . Find tan  = 48 – 75

36 +100
measure of acute angle between the

regression lines tan  =  27

136
SOLUTION :
tan  = 27
GIVEN : Y ON X : 4x – 3y – 120 = 0
136
x 2 = 9
1
y 2 25  = tan 27
136
STEP 1

Y ON X : 4x – 3y – 120 = 0
X TRA SU M
3y = 4x  120
b ri ng ou t the i nco nsi s ten cy i f any i n the
y = 4x  120
3 3 f oll ow i ng

a ) b yx + bx y = 1 .3 0 a nd r = 0 .7 5
byx = 4
3 b ) b yx = b xy = 1 .5 0 a nd r =  0 .9

c) b yx = 1 .9 a nd b xy =  0 .2 5
STEP 2
byx = r . y d ) b yx = 2 .6 a nd b xy = 1
x
2 .6

4 = r x 5
3 3 a ) b yx + bx y = 1 .3 0 = 0 .6 5
2 2
r = 4
5 b yx + bx y  r i s no t sa ti s fi ed
2
STEP 3 he nce i ncon si s ten t
bxy = r . x
y b ) b yx , bx y & r mu s t ha ve the same si gn

The ab ove co ndi ti on i s no t sa ti s fi ed and

= 4 x 3
5 5 he nce i ncon si s ten t

= 12
c) b yx , bx y & r mu s t ha ve the same si gn
25
The ab ove co ndi ti on i s no t sa ti s fi ed and
STEP 4 he nce i ncon si s ten t

m 1 = byx = 4
3 d ) s ta teme n t i s con si s ten t

m2 = 1 = 25
bxy 12

STEP 5

tan  = m1 – m2
1 + m 1 .m 2

- 26 -