0 ratings0% found this document useful (0 votes)
41 views32 pagesCorrelation Regression
Correlation and regression
Uploaded by
voxbldramaCopyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
0 ratings0% found this document useful (0 votes)
41 views32 pagesCorrelation Regression
Correlation and regression
Uploaded by
voxbldramaCopyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
You are on page 1/ 32
14
yields of crops,
‘Mathematics a the B.Sc. examination maximum marks of
jects being 50.
oll Nos. FIEIED
“Marks in Statiatics|34]37 36
‘Marksin Maths. |37[37|34
ppp ap]
3539] 33 [36 (38 | 43
‘of,y, and for large values of x to be
associated with lage values of, Morcover the general trend ofthe dots is
that ofa straight line,
14-3, Bivariate Population
‘every member of which bears one ofthe valucs of cach of|
id to be bivariate,
Frequency Distribution
that in the whole record there may be more than once
w number of measurements becomes large, some suitable
adopted for the measurement of such variables. A table
ed from the seatter-diagram by sub size 6-dividing the coor-
nfo equal rectangular compartmet chin
iment the number of dots which fall
‘given above if we take intervals from 25 to 30, 30 to 35, 35 to 40, 40 t0.45,
45t0 50 fora and the same for, then the dots may be arranged in atabular
form given below:
ye 5-30 | 30-35 ] 35-40 | 40-45 | 45-50
3-30 T
30-35 2 1
35-40 1é 2 4
40-45 3 1
45-50 482 | Mathematica
___Ttistobe noted that if any dot falls on the margin point
in the néxt interval. A table of this type is called a correlation table,
___ When an increase (or de
increase (or decrease) in the:
is negative when inereasc in
vice-versa.
s/ 14-7. Karl Pearson's Coeffictent of Correlation
The coefficient of correlation, r, between two variables X and ¥ is
where x,y arc the deviations.
{2(y)/n}, anda, and a, being standar
Conetation and Regression | 443
3. The students got the following percentage of marks i
id Statistics
Y_[y-¥-¥| 7 a
ap
w{ 7 |
nf oar]oa
a] <3 | 8
of ca] oa
s| os | x
‘Lo |
% es
wees poe
Rall Nas TTT 67
[i apart |
[star Ses [os [ee] BY SE | ST
a
jon sub size Gets be denoted ty X and Y
i Then the ican for X maths = $2 = 65 and ne mean for Yaris =
$2 = 6.x and ave the devon of Xan YS from hi epee mean
then the data maybe arranged in the foloing form:
x Y x ? =» |» |
eae
w|i] -29 | <1 | om] 2s | 3s
oe [91 | 33 | os | tas | 62s | Bes
' 2 | a |= | -6 | 100 | 6 | 20
| -@ | ow |. 2 | mo fo 4 2
a |-@2 | v7 | -4] 2 | 6 | 8
wo |e | 2s | me | 6s | sto | som
a 88 -3 | -8 9 64 2
ao | 3 ofa | oj w]e fy
fea | 26 | -19 | oe | sr | ast
750 | 660 a - Sea
we Ge
v=
es
| re X~¥eXx-65y=¥-Poy—o6
Mere arts $198, 29? = 2224, ny = 2704,
4 rs 2 a0
7 Vxq9 x97 V5398 x 2224 axa
= 2108 270400
Feng = 0-78.
tx-
ere ds = M,— Ay= EE
— Ay = Y= My +My A, = y+ dy
Met Me A, = + dy
B= Zoe+ d+)
= Ey tad, + ye + ded)
ent of correlation between the values of Nand ¥.
3 7 s 0
T 17 © a
jandeikband Bo; Tadure 90; Swat
npue 98; Raipur 90; Rohilkand 94; Vikram 92
Conetation and Regression | 445
for X'be 7nd for Ye 15,
e
7
2
a }
o
3
10 20 3 15
= [= 7 |
Substituting the wove values in Kael Pearson's Formula, we get
Un - Lz ey
V(66~ 10-67) x6 I
of Vearson's coeffciens of correlnion for the
wow
57]
in your o»
a
wp oT Tr |
ak 446 | Mathiemasical Statistics
Forcheck, 2¢ +1)=2@)4N =04 10=10,
E+ 1)= EQ) +N = -104 10 =0,
EG +1)'= 26) + 2 @) +N = 170404 10= 180,
2 (q+! = Ley} + 25 (7) +N = 150-204 10 = 140,
LE +VO+ N= EG) +20) +20) +N
=%+0~ 10+ 10292,
ulation ofthe chart is eorect.
the value in Karl Pearson's formula__
ay ~ BEL.
=
- [eo bea
0
J Yar 0,50 10)
eM:
2.
= = =06
WV 7x14 OXI
Remark: Letx andy be two variables Let Mand M, be their means, then
ze
= y=
2 —M)Y = M5)
Cov (
alay my
etay-tots,
vane =2zq2at= tor (12)
1 i»)
Example 3. Calculate the Karl Pearson's coefficient of correlation benween X
fand ¥ series
Correlation and Regression | 447
Solution = Correct vale2 of Katt Pearson correlation coefficients given by
vy byery
fe-enh
dinensiivii=n eniywnusate
sth, , 448 | Mathematical Statistics
= 3-4 +6) 412 +8) ay
= By X14 +86) + x12 +6 xR)
227 7 Gt +48) + 06 +40
[0B =
520 ~ $00
“e}
(650 ~ 625)(936 ~ 400)
20,
V25 x 536
20.
= syn es 2017
Remark : Ife take 460 in place of 960, then
ihe coefficient of correlation for the following table
Conelation and Regression | 449
12-
3-3
fot Lenn! fer an
1 78 z 26 30
@ 2 6 30) 36
* = 7
=T 3 7 T
2] 4 4 4
er) 1 °
-3 | 0 ° 9
1 1 1 1
2] 4 4 4
= w om
og the deviations of from 22 and then
ing the deviations of y from 26 and then
ST
26 Eo a 35
a
assumed mean for X and ¥ series bel$ hnd 29 respectively
E=X-9 [a=¥=29 | En e =
= a
=a] 2] ‘ef ue] 4
21a} 6] wy 3
of o! of} of &
—— wo} st so] m | ox
; re ee
den'|fa*— Zan} Sais 450 | Mathematical Staisiics
Conelaion and Regression |
egity =o,
fice of correlation bemcen i heghg
comin oe flog se 0
re Meo cnineae ft
=e Sse
a0 | Ta |
Ee an | =
I Fee 7
| Teg pa Say
a] %
ra)
ad
ol 4
40
> 4
3B 1
si 4
1a | 16
| 20}
pe- fn | i
|
I
}
i
+
{7 x 542 ~ (46)°}(7 x 358 — (6)')
. 1197 - 26
Yea794 — 2116,2506 ~ 36)
tat oa
© Vi68 x 2470 XT © STS =
ple 8, Show that the value ofthe concen of emelaton i indspndert
of reference and the scale of reference.
J With usual notauon we have
~)
lependent ofthe change of origin and sie
yk <00rh <0, k>0
tions ofthe variables andy fom tit
Conelation and Regression | 453
P's rg
\ aes
ate]
ar Fat
= alone 2a) =o ens 2m
=ler
uy
ies
BEG +B) ei.
ea) “
we have
BED
Tuner LILS. of (1 is iva
quantities
* lsr20
* i4r20 and d=re0
- re-1 md rst
atsrst,
Apezample 7. Prove tha the Pearson's coefficient of eoreaton + tes
' bebveen ~1.and +1.
Le waz -or- oF Tashumnty,
cous rinimumit
Be ~22ey-ar~b) 20 {Minivan
Ha - 2G -ar-8)=0.
andiherefore = a= 29.
La
8)
negative being the sum of squares of real
Since dy =O= de,we gerd = 0,
3 & volat 62
w
usry-ayBy'-Udyrane & fF
29-227. ny4 ‘,
wane (Gt)
ay! 20M, Gor oy? xt
we a
fo 454 | Mathematical Statistics
say,
sng tenn eager
‘Simviary, 24%, will not be negative. Hence.
1-F20 ost
‘pvingthelimic ofr as~1 ond + 1
EXERCISE14
Sw
oe}
~». DF
¥
2
*
10
ry
ietacen +
Beweda~ Sard +75
Retween Gand 4°5
o
J Beaver —
Detweea = '75and ~
Tween ~'S ard
| °
rT)
0s
an
485
sta
a5
19 450,
197 m
199 ss
a
sw
ion Coeficen. Show thay Karl Pearson ‘Coretation and Regression | 457
xa
fat 3731) (ay
(8-23) (7-2)
s 2736. 308.
T68
1 Calle coefficient of oeltin frie loving ble: atu ore me coven x a et —12 forts
0-4 4-8 8-12 12-16
T | |
6 8 |
3 | Tx + {
7 2. | = 20inthe row = ~ 1 isobtained as
9 2) + 8(-1)
< D ) 'v = Ois obtained as
Sg 3x0.
G _% rowy = Lisoblained as
+200,
veya row v = 2is obtained as
a Simian ie oor ebm )
| | “naieciumn =~ 2ecesbesss (bv
h t
Find out fu, fs, fi? and fv?, Now second column from th
= 14in the rowy = ~2isobtained as
) = -20ere.
fo byw and fa by v
0 |
ay
. =e J
|
:)
r=
+
x74 (exe #8) |
+o+ Chinas) 458 | Mathematical Statistics
Comelation and Regres
16-18
18-20
2-22 ical Statistics
Cometation and Regression | 461
96:5 | 98.5 ;
100-5 | 102-5 | 104-5] 106-5] 1083 sno EXERCISE 14 (B)
il3t el: Trt 7 A tind Kart Pearso's oelccnt of corelation from the following tbl
1)2/ 8/2 Sye}zfafa Tos [5-0 [0-5 | 5-0 | we
ofa i ot T z
s}o}wfo]o ile a8 4 s 5
3} s]e]a 1 ai 3 ‘
4}2}3a]4 12-16 2 1
afa 1
ada Fa the coeticient of crtelstin rom the following able:
o-4 8 e2 B15
T
2 ‘ 1 2
5 3
. 4
2
mn between andy for the following tl
sae ar] |
$ Sep 3
3 ew | - | = | »
wo} - | w | wo | 7] = | 2
wo} = | - 7 | o | « | a
sw] - | - | - | «Js *
Tout_[ 9 a9
i, )Gaesate the elfen of eonrationbaween x ay
Y x 0-8 Bo1G o-4 wu-3
af =H v5 7 =|
=e 5-10 6 *
w-15 5 3
15-20 7 2
n-% °
rs +
5 Calculate the coffe of colton fom the flowing tbl ging the apes of
esbands an hires in years
Agcot | geo bane Ton
Wee [SoH] RO | 0-9 | OO | Ow
Bs] s 73 7
35-35 wo | as | 2 ”
SAM s36 ~ as} 35-45 1 2 2 5
i 15-55 af} ow | os | a
t eS, _ 5 55-65 | 4 2 6
Viaovio x sav20 | Y10736200800 “Tea wpe} 482 | Methen
atsties
“he fotowng tle gre sng to ae est of mse
Fleer or aaa
ae
Mats Fy es x ‘Toul
eT 7
awl s] § ‘ 2 |»
nw} 6] 8 » | vu [os
ws] a] 4 ‘ e | 2
9-0 2 4 a fo
an 2 3 1 ‘
Tat | | 2 3 |e
(Cie the osthecns of coreon beineen apa igen,
CClevat the eoefcien of corcation for the following abe:
Bs [Hs] Bs [ws a5 [TS
Conelation and Regs
i@ 14-18, Standard Error and Probable Error
‘The formula for calculating standard error is
1?
See
4
where isthe coefficient gies
mans ae equal cach being
142+. (ut
tauth
> k= BYE = 2
La d=X-¥= ( ah. (2H)
from the meat
——
waz ( Ay. nyt yxt8 (2 ical Statistics
= n+ (Qn +1) _ (n+) nt)
6
| 468 | Mathem
Similarly,
syn nD
Now from (1), we get
Xd? = e—y)' = Ex + Fy - 2B
= By=}@e+ yt Ea)
a fio D — sal = yn =3y~ dat
Putting these values, we get
This is the required formula for the coefficient of correlation of ranks
is duc to Spearman.
|. This method
ranks, Ineach series the
largest 2, and soon,
ues
inimum if Za? is maximum ii
‘maximum, d will be maximum if the ranks of the 1
iowa below :
3
a2
odd and'whon a b crea
odd and equal to 2 + 1. Then different d's arc
(BET = 1), FETA = 2)y oy 4, 2,0, 2, 4 ony “(P= 2),
Correlation and Regression | 465
fe, 24, B= 22 = 4, yO = B= 4 aig = (= 2),
Ed? = 2 (2) + (2-2) + + P+}
= 81 +¢- 1+. +p = EEN +Y
6
6rd
aw)
a xd"
rang grey cin ae nk
Case2. When n is even say 2, Then the values of d are :
= 1), (23), ony y= 1, -3; on (= 3), == 1)
Bd? = 2 ((2r-1)'+ (r= 3) +. +1}
= HQ) + (= 8+ 2+ PHP LY
{+ (2) + ae DY
QPP HF ++ W424 Fe- 1 +.. +2")
= 2[52r(2r + (Ar + I) br(r + IY + 1) = 30 4" —
GE? a (4? = 1
roi-—Gie -£e-» ~
net) ~~ ar a1)
‘Thus we get ~ 1S rs 1 which are the required.
ids for the
ced; = O for all values of i, therefore Zd° = 0,
Henge r= 1 ~ 6 PE a,
aksandX,+ Y, = n + forall
Min- Y= dandy + ¥oent1
d= 2X, - (n+),
2d? = 4EX' = 4 $1) BX tn (n+
. AVQ+D ging need)
+1
and (Correlation and Regression | «
Staple 3. Clela the cafe of comclaon from te dara give
Fie tet harmo ifr: Mae ey
5
Ts Lise |e | tor | 86
7 i 4 é Solin:
3 6 = x y | Rankin | Rankin bra :
5 4 1 x ’ 5 é
8 9 -1 , e a4 7
4 8 - » 2 | 2 0
1 1 6 a 1 1 0
0 2 8 8 5 : 1
* 3 i » 3 3 0
a ” 6 6 5 1
6 5 1 2 8 1 1
9 1 Taal ae a
Bal {
Gx4 1a t- pet no
r= oS area
s1-13=-05.
Fee, Te competitors in a beauy contest got marks by tre jst
3 es
< K ... Example 2. Find the rank correlation ficient jon the following data :
. x 10 7 iB wd
¥ a a ao
ee ak
ee
%* Y Lr ail TERT ges-y e
5 5 o a
4 4 0 °
2 % “1 1
3 4 2 4
1 2 “1 1
v €
- oe
Oey oes
x6 x se
et Oka PER eH HOT
leggy Sx ‘or frst and second judge's opinion xa? = 200,
oxd? 6x20 7
Hence = 1 = PPE = 1 antag =
(2) For first and third judge's epinion x4? = 60,
ice thereis less difference in their approach,
sceond and third judge’s opi 2a = 214.
ic. there is much difference in their approach,
‘Thus we conclude that the (1) and (3) pairs of judges hgve much ds
their approach to Beaay.
/14-19, Tied Ranks
When two or more observations are equal in a series we. say that there.
is atic, I ties oceur, they may be assigned the average of the ranks they
| would have received if they had differcd. For example,
| MOF If two items are tied for third rank, then they may be ranked
3 oe Be
|
19
e
Ai) IF tree items are tied for 7th rank, then they may be ranked
T#S*® ~ 8 and the next is assigned the rank 10,
xi again there is some tie the same procedure i
In case of tied ranks, when correlati
then
ra 1 She +7
n@=T
used.
where T; ;
= G52 (m} — my for tied ranks in x
— iY in
and T,= 752 (m}—m)) for tied ranks in ¥
‘ny= number of equal ranks.
In case of tied ranks when correction is made in variances and
covariances then
bee =n - (+
T) - 3a?
Correlation and Regression | 469
Solu
- Y Rank ora ianstY dandy e
| a 10 T 3 =7_|-a9
| 8 12 2 f -s 23
B 8 35 3 nis | 205
B 18 35 5 -15 | 225
0 8 5 5 ° o
68 2 6 2 4 16
oe 20 7 3 4 16
58 24 8 1 7 | 4
Total = = = a
62d? + 7 -m)
rei-
met iy
Here m, = 2, since the rank of 73 is the average of two numbers3 and 4
6 {159-5 +, B-2))
Hence ee)
=1-27-0405xe _ | 46a
“exes = SOY
ot- BB a1 -192-05. = 0-4
Fxample 2. Calculate the c ficient of correlation from the following data :
A [4s [36 [58 7 45 [a0 36] co | 30 | ae
Pee ao 36 ae sae aoe]
¥ y [REET BREST =
40 Se 4 TS
36 25 6 “35
30 8 9 -1
4 4 2 2
36 5-5 6 -05
32 7 8 =
4s 25 1 1s
a2 1 3 22
20 10 10 0
36, 9 6 3
— 1 — = [=o 7 470 | Mathematical Statistics
In the case of ticd ranks,
Gea + BE omy} —m))
a7 oD ma)
Here m= 2.since the rank of 56 is average of two numbers 2 and 3.
m ce the rank 45 is average of two numbers 5 and 6.
since the rank of 36 is average of thrce numbers 5, 6 and 7,
(nm) — m)= 7x (@ - 2) + 2-2) +e 3}
wtheeeeny = 3a, ao
‘Hence rank correlation coefficient is given by
{from (1) and (2))
=1--8136+3) 2, _ 6x39
Jodo =)” 90
= 1-0-2364 = 0-7636, °
EXERCISE 14 (C)
1, Two Jadgesin a beauty contest rank the ten competitorsin the f
e[s,73]7tl?]7]°)75] 0] 5
apa. bep7ts teste p ets
dards?
lowing data, showing ranks of stu-
‘denis in two sub sine Gets:
Mathemais] 3 Pe] 2 ][2]7])o]* |e] 1
Pays spstepiy,® a p42
13. Ten studeats got the following percentage of marks in Uzonomiks and
‘Stodents TIl?]72]4]5]¢]7]*]91©
Maris in Eeonomics es [Ble | | alo |»
‘Maris in Saisie al@l@lale|s|s| a7
(Calculate ihe rank correlation coef:cient.
4. Caleulte rank correlation coetficies
(6, Marksof twelve studentsin Arithmetic and Algebra are given
Frimener [O [MIO DSO zlSlel sf alc
s[n [a ols[sl2i~ls[elals
Catzulsle the rank correlation coe
ent betweeax andy.
Mars [1] 2) 2] 415] *]?
Pres | [103 [4] 5 [712
the rank correlption coefficient for proficienciesof this group
Siudent ALB
Rankin FietTest [$5 [5:5 |
RankinSecond Tea | 3 | 5
9. The ranks of same 15 students in
‘herearenetmoethang (? =n) 1 pouibe vauete
18, Define nk conlaton andere the formula
14, Show thatthe eoef
15. Define Karl Pearson's coefficient
correlation. Also menti
between ~ Land + 1.
ion ies between ~ Land +1
orrlation and Spearman's rank ete
roperies. Sh a corcationean®
14-21. Regression
im ‘repression’ was first used by a British Bi
Francis Galton (1822-1911) ic later part of nineteenth centr)
connection with height of parents and their offeprings. He publi
Paper entitled ‘Regression towards Mediocrity in Hereditary Stature’ © Correlation and Regression | 473°
yractb
wherey = ¥— M,andx = X= MM,
ifference between
point and the
ofs
U=xy
ofr,
Out Weight oF vice-versa consider
1c former,
old one we get
Yom, <1 Bara)
the mean of y-series.
means M, and M,,
ofx. Changing 474 | Mathematical Statistics
stature of sons of tall parents reverts or regtesses back towards,
the mean stature of the population.
Note2, Ifr = +1 oF —1, the two regression lines will
ly negative)
values of the other. Ifr = + 1, variables are perfe:
tively correlated, high values of one corresponding to kigh
values of the oth
Note3. Ifr = 0, the two lines of r
1M, which are two lines par
their means M; and M,, They are perpendicular to cach other. It
means that mean values of X and ¥-do not change with ¥ and X
respectively ie. X and ¥ are independent.
Note 4, Following are the diagrams for various values ofr:
sssion become X = M, and ¥ =
°
YO 0) (aco (ea+t)
_AG-25, Regression Coefficients
‘The coefficient of x in the regression line of y onx
coofficient of y onx and is usually denoted by “ts.” thus
be =r
‘The coefficient of y in the regression li
coefficient ofx ony and is usually denoted
bya rZ.
3
14-26, Properties of Regression Coefficients
1. Coefficient of corelation és the geometric mean of the coefficients of
regression.
lled regression
ny is called regression
thus
‘We know that regression coefficient of y onxis
and regression coefficient of x ony is
3 oe
bye re
a
+ Geometric mean coefficient of by, and be, is
Jicig = Vidixr® «VF =n
Remark 1. The sign of rdefined on the sign of cov (x,y).
LIV Gov (&,y) > 0, then dy, bry, 7 > 0.
In cov (x,y) < 0, then by, 6;,, 7 <0.
2 tic Mean of the coefficients of regression is greater than the
coefficient of correlation.
We know that the coefficient of regression of yonxis
byw
a
and the coctficient of regrestion of x ony is
‘Then arithmetic mean of b, « on b:
bya tbe, (0)
it ilaed | > r
Gp tos > 22.0,
(, — 2,)" = Owhich is truc.
be ir =#
var (y) = var (v) 1 Fens 3; Tie folowing marks have ben obtained by aelass of students in
j «ov te cin te) ies (oto
| Now ban = SL) HE he, twert Tas [35 [36 [SS [OLS SDL ELMS
| qe @) Papert [36 | 30 | 48 | 60 | 62 | 6 | os | 70 | 74] a2} oo
i: Wiute the coefficient of correlation or te above data, Find also the equation
i bye = Ree ‘Of regression. ultr),, ree * cme
| . “y . lon: Thking A: = 65 and A = 7Oasassums we get the follow
Dy vee corn | SOE, Taking assumed means, we get the following
| hw aa ) Tag) wer (___—PaFERT ———]-——
i # x é e y me | oF
i . a
i 1 100 | 50 |
| From (4) and (5), 56 “9 ‘81 48 it hs
i pendent of the change 58 7 49 60. ~10 100 |
| g “3 3 | ot ~# 64
| © ° 0 | 64 -6 36
\ s 3 9 | 6s -s 2s
i 5 r ‘0 ® :
i 100 (s 4 16
a0 i | ms | aw 2
85 20, a
= 400 20 20 400
49 THES.
Formula, we get
Jes whch tne two regression lines make wih he
tans rh tan = . 478 | Mathematical Statistics
- pe BB BH [
# Fr y
= ) 0 i | 2s] 8
35550 x 18218 a | -20 | 400 | 56
Regression coetficient of xony 55 -10 100 50
eho) 2x (-49 sx | -9 | a | 4
so | -7 | 4 | oo
@ | -s |] | w
wo 0 oO 64
6 3 5 | 6
0 s | 2 | 7
75 10 | 10 | 7%
8s 2 | soo | 9
az | wey
‘Karl Pearsons correlatio
1393 - 2xcs
om pare
aL ma «Gi = 6s.ss mans -
(2 = 499°
{usta — F2} frsos — 25)
{on OT THe Tie of regression of ¥ on yis
Mean of U
‘The equat
ie x65 2=0-85(y - 65-55) ~
a= 859 +948 (2) ;
Sins the equation of he ine of regression ofy ons
y — 65-55 = 0-99 (« — 65:2)
. y= 996+ 8)
fe 2. The following marks have been obtained by a class of students in
‘of 100) = <
oo [oe] | mls l= Fo VES OT
Papert | 80] 45 | 55 | 56 | 58
a | oo | 62 | 6
‘above data. Findatso the equation
65 (7 } 74 | 90
Papert | 82 | 56 | 50
‘Compute te coefficient of correlation for tie
of the lines of regression.
Solution : Let A, = 65,A, = 70. 480 | Mathentical Statistics
= 19-698 = 12.235,
Resvession coefficient of y on.x
Byer S = 0919) 42-235) _ 21-248
GR) = 3G = 0-992
Regression coefficient of xn y
(919) 01-336) _ 10-418 _
2 = Bs = 0-851
Again = 65-2 marks
Peay + 3h = 70+ S82 = 65.55 maris.
Equation of line of regression of y on xis
d-ParBe-H
y~ 65-55 = 0:99q — 65-2)
Y= 0-98 + 1.002
ion ofline of regression of x ony is
e h
Obtain the twa lines of regression and calculate the expected average
fers Scns and he expected aver
= [9 =
=a fF
a3 | a
2 5
2} o-t
at 3
a 1
9
4
4
1
° 1} 0
4
6
3
1
= | ae
2g B= 69 -§ = 0835.
Arithmetic Mean forx series. ¥
1.
Correlation and Regression | 481
=i (B)' 61075 erage. 89.
Arithmetic mean fory series,
FA +B 09-12 68:5,
Correlation coefficient, r=
ro a3
Vs4—Hee2a- 2) VBx40
a2
47.
Vis
ence the line of regression of yon:
yo MarZe- ay
fae yoese a7 xP Be ops)
or ye A212 +39-77 ~()
“Theline of regression ofx ony:
xa M=rZg-M)
ies ¥=68-25= 47 x FG 68.5)
5S
ea say +3908. -@)
ence ecresponing ox = 675i tw get
7-605 47 PB ors ens)
« y2685 + 1923 o5) 2 69.5— 1m ces9,
Thus the expeied rag eit oh sons 68-19 inches corresponingo
the even eign ofthe faker =
expected average eight ofthe fathers
Size S2extass +3229 = 70 a0inehs from)
ale 2 Find he vo linc of reps an cuefinne of con ann ee
Ti dan gen Selon ;
= 18, Ee = 12, By = 18, 5:7 = 60, 5)? = 96, By = 48, 482 | Mathematical Statistics
60 12)?
=. (if)
3-33-45 = 2:88
= VERB = 1-7 a
rey
(nx? — xy Finzy? = Gy)
18 x48 ~ 1218
[18 x 60 ~ 125018 x 96 — 18)
os Gis
298 _ (18)?
=it~ (ie)
3533-15 4:33
= VSB = 2-08
864 — 216
(1080 ~ 144)(1728 ~ 324)
= 88 __ es. 30-56
Voss x 1404 3059x3747
Regression equation of y onx:
y-FarZe-H
2-08
- yo t= -56x 2b 67
- y~ 120-683 0-46
‘= y= 0-G8x 40:54,
- Regression equation ofx ony:
ero
* £0.67 = 0:56 x p91)
2 . x 0-67 = 0-46y - 0:46
x 0-46) +021,
‘Example 8. Given the foltowing information +
‘Mean Bianeard Deviation
Yield of Wheat AD a
itogram per unit area)
* [Raiiall (em) Z
Correlation coefficient between production (yield) and rainfallr = 0°5.
Estimate the yield when rainfall is9 cm.
Solution Le, z=yield, Y=
‘Then ¥=10, = 8 ond r=0:5
o = 8, 22
Regression equation of on yi given by
x-Ferky-j
2 0205x808
- x-10=2y-16
-- x= -6
Required estimated yield wheny = 915
ol)
= 2x9 - 6m 18-6 = 12 de, 12kg por unit area,
Correlation and Regression | 483
le 6 In a panially desiroyed laboratory record of fan analysis of
se ietelbuigtiadtonaregbies 7 2% 14 anala of
Variance ofx =9
Regression Guations 8x ~ Oy + 65 = 0, 40x ~ 18y = 214
What were (8) the mean values ofx and y, (b)the standard deviation of, and
(6 the coefficient of corelation between x and y
‘Solution : We kn6w thatthe two regression lines pass through the point (xy)
So to find andy we solve the given equation
&x- 10y + 66=0
~ 40x- 18y- 214=0
(1) x5 + 40x ~ 50y + 330=0
Subtracting (2) from (3), we get
yy-su=0 y= ai7
Substituting this value of in equation (1), we get
8x- 10% 17 +66=0 :
- ge= 66 +170= 108. ro Bhes
Thus #213,p=17 4)
“The equations of regression lines can be put inthe form
y=-tr+6-6, yonx
and x = “4sy +5-35, ony
<2 The regression coetiient ofy on:ris
rh n08 “0
“The regressitr coefficient ofx on yis
% = 0.45, a)
Multiplying (5) and (6), = 0-45 x 0-8 giving r = 0-6
Varianee of x = 0,’ = 9, hence 0. = 3.
Putting the values of and o in (5)we get, = 4.
Sor Brample2 Gens = 4y +59 i+ Saretoregeson lines ofxonget
Show that 05 4k 21, If k = 4, find the mean of
coefficient of correlation.
wa variables and
Hon : Given
x=4y+5 xony
yske+4 —yone
Regression coefficient ofxony, bay = 4
Regression coefficient ofyonx, by, =k
Hence bay bye 1
° thet Correlation and Regression | 435
484 | Mathematical Statistics
Agiinifb,, 20,then b,.20%k 20:8 4k 20 . fae
ey ethno Example 8 olin able gves he ag of hands and wives for SO nly
© maried cope: Fie wonaesontn asennad
Wonk = ten wenn ond aga ico akc
aan Aol Hasbant
Amore 20-25 25-30 30-35
ia 7 a =
| 20-24 6 u 3
Hi 1 24-28 _ 7
|The regression equation fork = fare
seares)
| yaad
‘Foup of husbands
aa a mwa oa proper ene
Sova tese eqns we go = 28 = 3
ar -
sample 8. For nwo random variables x and y with the same mean the two rae f fe fi fs
_Fegesion tones sae ban 9 Son a a )apet
| pire 0 » {of o | o
| Find also the conmo,
1 7 7 ¥ 7
Solution: Given fs
year+b, yonz 0) 7 30 | =16 | 30)
xeay +f, xony @) =
brza and by =a
| common mean be m then 3
| Y-J= bj @-H ey-m=ae-m) © ‘iw 9 0 7)
=ax+m(1-a) : x J
Hwa! gah Hern oxy=m Meanots, reap Baus S5CD a5 -.5 < 2790
ym — (8) /
* ae Fab RE 2 2 4 SIO 99 1.96 = 20.78 900
f ‘Comparing (1), (2), (3) and (4) we get zy Bp 222+ SETI = 22 1.25 = 20-78years
i b= m( a) Regression cvefficiem oty on
| P= in ~2) pv = 23h
‘bm ~a) _ t-a p= UE) _ K savy)
. fe fo ete = are) TT ar eee a et
ail -
‘Again from (), ™=T-a | commonmeani=7 =m. ee
a poet =f
‘and from (6), ato toe 3 es
i Since regression line passes though @% 7) we have from (1) and (2) 25 - Ge
arin =am+b hI 16 4 ag
maam +B ES BISA * Bey AOA
amebrew +2
: am-am=p-b
° 486 | Mathematical Statistics
Regression coefficient afx ony:
Dyy = GHP) =H cov
"= arg) = R var@)
5 uw lt
= OO
“ 5 M4 com
=F W512 7" Bs ~
ence regression equation ofy ons
Y ~ P= bys (-X)
27a OE 2) w yn O-4Ts = 8.00 a)
Represion cquatiohony :
Set be-D
ta 4-27 20-12 - 20-78)
> rg = 0-T2y + 12-04 (2)
Enated age ofhustnd,
a BP 0.720 20.5 1208 {fom @}
2 deed years
») Estimated aot wile,
ee Se 0-47 230+ 8-09 {trom
= 22-19 yeas
Correlation and Regression | g|
the two regression ines and the coffin of contin trom the Eng
3 3 7 e 7
3 4 3 2 7
mW] Bayo
30 6 | 50 3 [a]
8 Find equations of grton nits
and Wives at Marriage: ”
Hanada [Ta [@lele[ o>, als > Spa
Wiles age CE Ee
form ofthe fllowang age of
usb
ng marks have been obtained clas of tudentsin Statics (ut of 100):
o[s[s[s[s|o|Gla@[m sa]
a[s[olelolalalelm>x|o
Poper
EXERCISE 14 (D)
{on coeffsient and the equation of regression lines forthe followting,
(Compute the cooticient of corre above dota
the lines of repre
10. The following table gives the various values of variables. andy
= T z 3 = s—_] @ ry
7 a
¥ z = 3 = oe | 3
‘i ine the repression equation
2. Fin he two regression fines fa th flowing a ae ris the repression equations wid
Ts 1s 17 4. aban) = oa
a Lele 2H ae ley 25 ying thee
En
18 costiient of correlation from the fllowing
4nd he mo eresion etn
wu a ee
Z x e 4 : 3s : Ts 12 Calculate the coefficient of correlation a
meets:
1c the methodology of computing the correlation
=li]?]3]¢
Pe
. ine of regression of y on.x by using the following.
Z z i
; ss 488 | Mathematical Statistics
Papalation Gn thousands)
No. of FV. seis demanded | 15[ a7_[ a7] 50
the demand of IV. sed fe
nts got the following marks in Stasis (c) and Mathematics ()
5 ES
CA A A
“Compute the correlation coefficient
te the marks in Mathematic of student who rectived 62 masks in Satisis
weekly
30 DH Dw] ole
esp So pes [OL Ss [sO as | 30
Calculate the regression line of sale on test score and estimate the probable weekly
sles volume if salesman makes a score of 0,
‘experiment 10 measure the stiffness of «spring the length ofthe spring wnder
Correlation and Regression | 489
21 Ina panially destroyed lboratory record of an analysis of correlation data, the
fovtoonng results ar legible variance of = 9, regression equations: 4x ~ 5y +33 = 0
22, The following regression equations have been ob
1 = O°S16c + 38°73, x = 0°51
the mean of and oso ofys aswel
the ratio ofthe standard deviation xt that of ?
ing data age given for marks in Paglsh and Mathematics inthe S.LLC.
two lines ofregresion and explain why there are two equations of regression,
fe the expected average marks in Maths. of candidates who received 50 marks
sh
Ay price in Bombay corresponding tothe price of Rs. 70 at Cal
lfferent loads ate measured as Tonbay.
Toad ings @) oF
‘Length in inches O): 25 L 35,
ice between to variables x and
ingeomtans
Zp = 100, Sy? = 460, Bay = 508.
ime of eheering that he had eopied dow 140
te corer value wore #1
ue
17. A comp
{yom 25 pairs of observations obtaine
= 135,20 =
(Obtain the corsect regression equations
© 48, twolincsofregression are given bys + 2) ~ 5 = Oand 2+ 3y ~8 = Oanday" = 12
‘Coleulste the mean value of «and y, variance of and the coefficient of correla
mnpur'
pur 90, 98, 6 Ke
{(b) the conrtation coeficient between x andy’
OB between the two prices ofthe commoditiesin the wo
the neigh (x) and weights (9) of 1000 policemen
nes, M, = 1501b4,04 = 2°Sinches, 2p = 201bs,¢ = 0-6.
rom the above dat
r ight of a particular policeman whose weight is 200 bs.
(b) the weight of particular potieman who isS feet
26, ‘The following data are given for marks in subjects and lina certain examination :
a o
icon Wore 3 =
Standard Deviation 0 k
of correlation between A and B = +066
inc the two equations of regression,
sponding to 75 marks bisined in
w the regression lines by using the
F=07 = 2G
estimatey whens = 25,
owing data
9 = 5,7 09
repression coetficiens.
tion coeffi
2 geometric mean of regression 490 | Mathematical Statistics
rithmeti mean of regrestioncoeiicents is greater than corielation cet
33, Lethe following statement correct ? Give reatons.
“The regression cocfiient ofx on yis 32 and that ofy ons +8"
3. I linear relionsip exists bewoon the vale andy, then prove tha he
toctclen bameen them ie Lor
36. Ike nes eesion oy on ands ony te reepecivey a+ buy tet = Oand
curt hay see Oprorehataibe 20h
36. The repeson ines Ly ons andsony ar sespecnelyy = ac + bande = oy +d
owe
Cmeonotas = RAE,
adh
and mesnofy,5 =
Tae
i) correlation coefficient bermeen x andy is Vaz.
‘37. Two variables x and y are conneeted by the relation ax + by +¢ = 0. Prove tha
coefficient of eorrclation betwoen them
ion equation ofy onxisy = ag
bo + biy, andr be the correlation cocficient
(sin = Aye! 1, where noanse,d > 0.06
40, Find the scute angle 9 between the two regression lines in ase of twovriables, Explain
the sgnifieance of @ when re,
of regresion incase of two varies.
sion lines. Find the regression lines for any given
and find the ange between them.
ofzand yrespeciively,
Solution : In termas of and v, r the coefficient af correlation is given by
reco
~ Year (u) var (»)
Hoe waxy sotham=z+5
and vex—y sothatver—s.
cov U4) = Ew — -%)
SEO -Rty-Nenz-y 4
=Ele-3'-9-9))
=E@-¥F-Ey-5F
zolgi.
Correlation and Regression
i
warty eB e+ yt EO + £0 +286)
evar @) +e g) + 2600 Gs
Simiary, _var(= varee =) = Var () + ¥9E 0) ~ 260)
ANS,” cov uy) = Caszandy ae uncorelaed variables.
varumyar@y + var )ie a= Valea?
Similarly, a Verre
Example 2. Tho rables x and y have means 5 and 10 ans
Sand 9 respectively. obtain inthe coefficient ofcorelation berween an
-20-)
= Dar (x) ~4 var) + 0,,y are independent
9K 4x 9=0
ov (44¥) =
oe
coefficient of correlation r between two variab
Example 3. Show sh
candy is given by :
02 + of ~e-y
1 Tae
where o2, g2.and a,—y?arethe variance of x,y and x ~ y respectively.
Solition: a -F =Ele-)-E@-
=Ele-»-@-7r
e-H-o-n
a2 + oy ~ ony
=e 7s Bor the following percentage of marks in mathematics
ete coefficient of covrelaion bythe method of diferencers
Markt :
‘Mathematics Sia
iB e
36 31
98 ~~ 1
2s 60
5 6
a2 o
90 86
o 58
65 53
39 a
X_= Marks in Mathetmaties, ¥ = Marks in
X | OY freak yey Fe] 2 eae
7%) [3 |e <5
36 | s1 | -29 | -15 | gat m4
9s | 91 | 33 | 25 | 1089 8
25 | 60 | -40 | -6 | 1600] 36] -35 | 34
| 68 | 10 2 | 10} «| 7 8
2 |e] 7 | -4 | mw] 6] 20 a
90 | 86 | 25 | an | 625} 400] 4 5
a@]ss} -3 | -s] 9] o] 4 5
6 | ss o | -13 } of wy} 2 B
39 | 47 | -26 | -19 | o76| 361] -s | -7
Go [660 | 0 | 0 _| S898 [ama | =70 o
P= 65,7266, 2=-1
ads S38 2 530.8, 67 = 2 = 200.4,
oP ony = A e214
alt ginany
tee
539-8 4224-4 — 221-4
“Da vea8 x VE
5408 _ $40.8
SINE = ET
= 0-78.
ie 5. Find the comelation coefficient between xand a =x.
Letu=a=x,then f=a—Z,
var (u)= E(u ~ i? = E @ - X= 0! (say)
cov W)= Ee - Mu-H} = -EG-H a=
Example 6, Let and y be wo independent variables with standard deviations
+ aid o, respectively Show thatthe correlation coefficient between x and x+y is
x+y ERE @+y) = Bi) +89) =245
var (e +y) = var (0) + vary) = 022 +02
(since xand,y are independent. cov &,y) = 0}
Hu 7}
“ml
5 Var (x) + cov (,y) = Var @) =a?
te pO ew
Vear evar) Ver(atta) Vasa?
Hed variates withthe same standard
1, show that the conelation coefficient
++ yanido be the standard deviation of the variable x ory.
Tar+y
var (u)= vat) + var Q) +2 cov (6,9) “
320 42d = 20 0+1)
OV Gx + 9) = cov (1)
Ele -Hu 7 =Ele-HlE-]+y-HI}
winowy +E ie DY -P) =o covey)
Bdd+n,
+. Corrlaton cote eben and + ys gen by
nee flea after
“Wage YE
Example 8. Tivo variates x and y have zero means, the same variance o* and zero
cotati Svat
wnceuartyikd en Pai eeeL jee
‘have the same variance # and zero correlation.
Solution :0,*= cos? a0, + sina o + 2sina eos a cov, y)
- =o (sin? + cosa) cov (e,y) =0, x,y being independent.
ad, similarly, 0,*= 0°,
cov (u,v) = E(u) v -¥)
SE (ose +ysina—Reota ~Fsina)
(eana-yeosa Zeina +F
SE (roose + ysin arsine - 008 2}, ore
=E@ cosatina) ~EG*inacse Pa cosa) 9)
= sine conae?sinacasac! + (n'a ~ costa) cove)
20
Hence coresion between u, vis 0.
Example 9. fu = ar + byand v = be ~ oj, where x andy are ineaswéd from
heir repectve means, u and » are uncorrelated, the coefficient of corelation
beowecn' and y is gen by the equation,
aaa
Bas
0-21 GONE =
wer@) cory]. g?
Joven) “mney | Fo
Bu
varus cov (use)) a eae pht] AEE corey]
ey EP =e fet
L vary cov tu
Laven Ne “ar
rg E+ Bas
_faruvarn!?
esas
Bet
Ey a
Alites. od=var Wy =a'o2 + bo? + 2ub cov (x,y)
Pe var) = Ba? + a} — 2ab cove»)
pov (u,v) = E {(ax + by(bx - ay)} \
Hab 02-9) + (6-2) covey) \
ado} = oot (uv) = (a? + bY} {ae a? - cov ey)
=e + borg! (0-84) )
= 9 Tet retald Fi.
cov (u, ¥5= 0.
EMCO Owe tthag Vier,
Example 10. fu = ax + by and v = ax ~ by,
from the means of ovo measurements by the same
‘correlation between x and y isp. If, v are uncorrelated,
49)= Unb 44, V1 =p
wand v are independent
represent deviauons
I. The coefficient of
thar
zs
Cometation and Regession |
setaton: Abin th sboe exp
a -ehate= leery aE
ae | rm cove 2a
welsh ef Dass
Example 11 Ifx1.%2%9 be ree uncorrelated variates having staderd a,
ions n,n, enrespetively. Obtain the coefficients of correlation between, sy
fata)
Solution : Let.
wax +mandy ex +5.
EW =E@it2) =E@) + Ee.
—E qi) + fa - EG}.
sotted, ascov (n,n) =0.
Similarly, varveo?t +o)".
Agin | Gov, y= EU - My - 7
SE(@-H+n-Be-h +H -H)
= Ele: ~ 2G -E) + HY + (e
= E [ea ~ 23)" + cov @y,x2) + cov (2, 7
sev) io
- Vigtealerra)
Meat (tt) var (v}
Remark sit oi? or? os? = o,say thear = 1
botimple 12. The variates xand y have varices oad respekt ante
correla with coeficient of corelaion p and & and y and defined by
5x 0080+ysinG,y =y cosd—xsin0.
Show that andy will be uncorrelated if tan 28 = a
Solution: Weave p= 2%) giving cov @.y) =p o12r
£m m= EG ~ Fen - 7)
= E [ccasé + ysin 0) ~ @cosé +7 sin 8)
{( cos @ —x sin 6) re:
ETE ~ Dy ~ 7) cos 9 ~ sin Ocos {(s- 2) ~ 6
SE[R-Dy -j)cos20
= 2828 cov 9) ~ fin 290-3. 496 | Mathematical Statistics
Now &,» wil be uncorrelated ifov ¢@,)'= Oe
tan 29= 22"). Zeman
etme ot
Example 13.2) are neo variates with variances a7 and o} respectvelyand ris
coefficient between them. Determiste the vaes ofthe consianisaand
re independent of r such thatx + ayandx + by are uncorrelated,
Solution: Let w=x+ay, vax ety
Ware, vaxt ey.
Since u, v are uncorrelated.
O=cov(u,v) = E(u - My - 7}
FE(G-D +09 -DHE-H +6 y-H}
= E@-H+@+H)E(E-Hy-) + ade y-7)*
=o2 + (a+b) cov (y) + 0b 5
Sor +0b53 + (a+ b)roG,
Since a and b are independent of r, we have *
@; + aba? =O'anda+b=0
an-oaS
Example 14. Find the standard error of estimate ofy and x respectively
Solution : The equation ofthe line of regression ofy onxis
y-yerke@-y.
Let (¢i,yi),# = 1,2, ...,m be the variate value pair occurring with freqiency f.
Let Yay+eZ@i-B,
IFS, isthe standard error of estimate of: then
SP HEA AY — yi? where N= 2F
=WUGrrZq-H-y)
ay a-no- py
=P e-B 0-7 Ee-H.-D}
BP Balto} = 2 Leov ey)
=P ai +a - 2. Bray
ao -P ep agta-r)
S=5 0-7
Simi nda of eximai of gen by
5, =a Py
Comelation and Regression | 497
Example 18, fllustrate by an example that two uncorrelated variables may not
be independe
Solution : Let
x] -3 [2 pt z 3
» 9 4 T a 4 a |
where y = ie, xand,y are dependent
or hence ray = Oe x end are uncor
measure of linear relationship.
x y Fd ¥ y
3 9 o ‘81 727
-2 4 4 16 -8
=1 1 1 1 -1
1 1 1 1 i
2 4 4 16 8
3 9 9 81 2
7 ra 7B 196 7
B
geting yeRo Bi
on @y=re-y-p
Loy sp
= R20 87
={-0xH#eo-o=0
1 alae (x)7_ 28 9 28
var @) = of = Fee 9) =3- =o 08s
eat 2
2 2 196 _ (28
vena ngze napa (By) =~ (B)
= 1176 * 784 _ 392
aa ag
reg
Veareavar0) 4/23 300
Te
Example 16.4 ruber of persons are measured for th
an chest expansions (2) and product mament conelation
lated. Prove that
3
Tay tye t tae 3,
-2
Soluton Ler uw =2=% ym: zz
2H then
E WY) = roe = Fay EW) Stow = Iya E (WM) = Pw > Poe
Eutvewy20 498 | Mathematical Statistics
EQ ty +084 Quy + aw + 20) 20
EQ) + EW) +E (we) + 2E (wv) + 26 (om) + 2 uz 0
THIF L42ryt Biyet ree? 0
342+ ht ragz 0
Tay +e thee 2B
Hoo
14-30. Moments of Bivariate Distribution
ctx, y be two variates and ket be the frequercy corresponding tothe
air (7). Then the moment," about the origin (0,0) of order rin und
siny is defined as
rel = Ef
where N = = fi.
‘Similarly the moment 4,, about the mean (5) is defined as
pany She Ow
Inparicdar pw = BS find
ba! = RE f=¥,
= RS fay
Bx
Hn = 0, pa = 0,
d= HE fe-DO-P
= WE fa -F pho FH
ben oF = WY - 3 = as —F
Ha = 6} = RES YJ) = wn’ -F.
Here wn is called the covariance.
14-31, Mlustrative Examples
‘Example 1. For a given bivariate distribution, find the straight line for which the
‘om of squares ofthe normal deviations is minimum.
Solution += the gen tne be
reve +ysing=p.
ines UaIeeora+y ina-pj
Corelation and Regression
‘Normal equations are
M 225 (coos + ysine ~p\-rsina +ycoseyeo
ie, = cos a sin a Eft? ~ 005 2a Bfiy + sin a cos a By?
+psinalfe~pcosesy =0
Again Ha 207ecora +ysne=p)=0
ie caethetsinety=py. |
IEX, ¥ are the coordinates of the mean, then
iE yo.
xf, yo
@),we nave
Xoset Ysine=p 4)
line passes through the mean of the distribution.
‘Now taking the mean as origin, and in that case noting ti
B= 0 = 35,
the equation (1) is reduced :0
= $sin2202 ~ cos Zap + bsin 220)7= 0
giving tan 2a = et BC}
ao
“Thus (3) and (4) give the values of and a therefore the required Ines
cbiained,
Example 2. Two variables wand v are made up ofthe sum ofa umber oft
asfollons:
U=ntRt tnt tnt th
uanin+oimtatnt ote
where n,k and | are all suffices and where x's, y's 2s are all uncorrelaed
standardised random variables. Show thatthe correlation coeficient benceen ee
Vers bn +
Show further that
X=VeaidesVashy
and ¥=VerDu-Veshy
are uncorrelated.
‘Solution : Since allx,y,z are uncorrelated standard variable, we have
E@=0=E)=£@)
V@QEE@)=1V 0) =EV) ALY =ER)=1
cov y= 0=COv Hz) = COVO,2) EU) = EH) =O.
Now cov(uy=E [tu -EW)}{v~£ OH =E
SEl@ tnt thtnent +H)
Gitnt tment
are) =e) =n
var eve ta te ty tat Hye) | $00 | Mathematical Statistics
= VR) + Ge) + ke) + vO + — + HEH,
sn. ‘covarinace tes Var
Simitay, var =n +l,
fr be the correlation coe
Now EQ) = Vas Buys Vanek Eo)=0
E()=Vn+1 Ea) Va+kEW)=0
cov Y= B [1 - BEQHY - £0), & Ea)
SE UVa tet Vine kot xn fa Vinee}
=@#NEW)-~ CH +H EO)
= +)o2-@skoe
ECHO +4) (+h =O.
stray LEY 9 ’
ease X, Yare uncorretated, °
EXERCISE 14 (E)
1. Landy be two independent variates with o, and 9, the standard deviation respectively,
prove that variance of ax + By is
uP Bok
ZAC xy,2 be independent variates cach having the same standard ovtioe, Uf
ax-tyandy = y-+4, prove thle oefiint of comelation betwen anv ts
ates with vaisnce 0 and. resposively ant isthe
etween them Itu = x+y = xt Sy, fed abe vate f
3. xand yare two independt
4. 10;sthe ange between the regression ines of twovaristes withenerutarcouiceat
1. pre that
sio®@ 1-7
S. Two indepedeat ‘variables x andy have ero means, variances «7 and ¢," respectively.
‘Show thatthe coreation coelfieat between twa vasiatesu ana ¥is
OG rea
where = xeosetysing,y a xsing yeas,
6 if andy follow she bivariate normal law wth sro means aad vasisace oy" ant o:*
‘eopecticl and vorelation coefficient p, show that
= weak
u