289 Analysis equation by taking deviations from the late ssion below calculate the regressioi si SX when Y is 1 id estimate Y when X is 12 and s[6[7[s8 3 | 9 [12 [15 [16 14 data giver series a” y _de-S) de? he 10) dy? dedy 4 3 9 6 % 18 6 2 4 4 16 6 8 =I 1 2 4 1 9 0 0 -l 1 0 2 1 1 2 4 2 1s 2 4 5 2B 10 16 3 9 6 % 18 n 0 B 12 38 ee = _ IX 35 _ xy 7 Xe ee We No 7 =10 x 1 > 4 o * xy < ' < gression equation X on Y : X= bxy = 5. bxy = Tp = 0.41 gression equation X on Y:X=X -X = bxWY- ¥) X-5 = 0.41 (Y-10)=X-5=041 Y-4.7 X = O4TY-4.7+5 X = 03+047Y en Y is 14 X =0.3 + 0.47(14)= 6.88 gression equation YonX:Y=Y -¥Y = byx(X - X) ty Ldxdy ~ Orbyx => ve Eas? 53 byx = B =2.07 Bression equation Yon X:Y=Y -¥ = byx(X - X) Y-10 = 207(X-5)=Y¥~-10=2.07X- 1035 Y = -10.35 + 10+2.07X Y = 0.35 + 2.07X ‘When X is 12 value of Y = 0.35 + 2.07 (12) = 25.19.Quantity, yortcut method deviation: calculated from ne 4 3s Shorteut Methe le i mean is not a whole number shortcut methoq Mey H not actual mean, When obtain regression equations, ? cession equation X on Y:X "X= X = bx¥- Y) Regression equation Yon X: Y= Y- Y = byx(X - X) NEdedy — (Ede Edy) N&de? = (Zar)? tal of products of deviations of X series multiplied withthe deviation, Say = Total of deviations of ¥ from assumed meg, Sar = Total of deviations of X from assumed meay oty byx = By? = Total of squared deviations of Y series : Lx? = Total of squared deviations of X series | X = Arithmetic Mean of X series ¥ = Arithmetic Mean of Y series N = Number of Items Nlustration: 5 The following gives the age and number of morning walkers. Determine to ‘equation and estimate number of walkers at the age of 55. Age Years X_| 20 | 30 | 40 | 50 | 60 | 70] 80 k No. of Walkers | y [| 4 [6] 3 | il 9 [44 i Solution: ¥ Y_ d(-50) dN) Pedy 20 4 -30—«900— 6 36 1830 30 6 20 4000 16 80 40 8 -10 100 4 20 50 10 0 0 0 0 0 60 9 10 1000-1 1 -10 70 4 2 «400 36-120 80 1 30. 900 +9 ) 350 42 Oo 2800, 28174120 _ NEdedy - (EadeyEay) Ned? ~ (ayy? 7X = 120 - (0-28) 7x 174 - (-28)? bxy———_—S analysis hy D 118 784 by > . Msc Mh arte on Qqualtor x y - X = pesre™ equation a X =X = byy- 7) = 50 © 194 (Y= 6)= X~ $01.94 Y + 11.64 X = LOY + 11.68 +50 X = 6164-14 earesion equation ¥ On X YoY -¥% = byx(X -X) bx = aa ? = (Ed)? 1X =120- (OX byx = x04 7 x 2800 - (0) _ Mo-0 oe ees =0 byx = ven = 0.043 pégression equation Y on X: y=Y-Y = byx(X-X) Y-6 Se bone Y 6 =-0.043X -2.15 Y = 2.15 +6-0.043X ¥ =3.85- 0.043X soma smote oak b= 3,85 — 0.043 (55) = 6.22 n: 6 Find the two regression ines by using the fllowing data and estimate ¥ when X is 9, 10 and 2 X15 | 18 Tate 25 [19 | 35 y {io | To Lag Liz] 4 | 4 Solution: oe ee ae ee 15 -13 9 117 ie it” 10% 10 6480 x 4 + eo ers pg otra 20" o 0 El 7 3 9 2 4 6 os» 2 £ hs 23 «1S 19S go Sle, 8 4 OR ee ee ee 235 319 1222 1528 8by sion ation X on Y: X= nen X-246 = x= X= egression équation yonX:Y= byx bx = by oy Régression équation Y on X : Y = Y-15.24 = Y= Y When X is 9 value Y When X is 10 value Y When X is 12 value Y Quantitative 7, ‘thoy = bx(Y- Y) eared) ~ (Bay) 8x 319 —(-27X-30) gx 272-(-30)" 2552-810 ~ 1276 va 1276 x -X = bo ¥) 1.37(Y¥- 15.25) = X - 24.6 = 1.37 Y - 20.89 137Y — 20.89 + 24.6 3.71 + 1L37¥ y-¥ = byxK-X) NEdedy - (Ede )(Edy) Nix? — (Edx)? X- 1.37 8x319 = (27-30) 8x 421 -(-27)° 2552 - 810 2639 1742 say = 04 2639 Y-¥ = bx(X -X) 0.66 (X - 24.6) = Y - 15.24 = 0.66X — 16.24 16.24 + 15.24 + 0.66X 0.996 + 0.66X 0.996 + 0.66 (9) = 6.94 0.996 + 0.66 (10) = 7.6 0.996 + 0.66 (12) = 8.92 sient of Correlation Caleulation of regression equations by using Karl Pearson's Coefic Régression équation X on Y:X=X-X = 1 (Y- ¥) Régression équation Y onX:Y¥=Y - ¥ = 2 (K- x) Zdrdy e _ Eddy Se x pt oy ~ Nexoy293 analysis io" tio session equations by usi Fp ate 0 Te ven i by using Karl Pearson's ‘ wae ete given information Estimate marks in ae Mee eee jaf in Economics are (sand Economies | X | 65 [5840 [6 6772 [4854 F 76 | 54_ 66 parks it warts 10 OT y | 70 [7s 62 [4s 78 [oo 40 [64 [4561 siti a ee pa ee x 5 3 10 > < a vida) dy Eidedy 7 Or," Nesey eae x? 1” Noxoy 371 > Fir30%1520 Regression équation X on Tian ~ 371 {717600 3 ¥* 73106 se - xn Ay 1130 ? oX = 9” = 10.6 = 0,28—s SE Semana 294 Quantitay, ative rg i - N 7 1520 123 ks 10 . Régression équation X on Y : Régression équation X on Y : Régression équation X on Y : Régression équation Y on X : Régression équation Y on X : X=X-X =1=(-¥) a ¥ 10.6 X = X-60 = 0.2853 (Y - 60) X = X-60 = 0.24 (Y-60) = X -60 = 0.24Y-14.4 X = 0.24Y- 14.4 +60 X = 45.6+0.24¥ Letty = Y=Y-¥ = 15 (X-X) 12.3 Y = Y-60 = 0.287>= (X-60) Y = Y-60 = 0.32(X-60) Régression équation Y on X : Y-60 = 032X-19.2 Y = 0.32X-19.2+60 Y = 40.8 +0.32XQuantitative Teoh 298 11. By using the following information obtain regression equation Y on X. and X on Y. ia | Standard Deviation __| 2.87 | 1.4 | (X = 1.99y — 14.9, Y = 8.2 + 0.47x) ression equation Y on X. and Xony. 12. By using the following information obtain regi Standard Deviation Co-efficient of Correlation (X =-6.4 + 0.95y, Y = 7.25 + 0.95x) 0.95 13. Construct regress equation for the information given below. [x [23 [ 23 | 53 [| 63 | 73 | 8 Ey |-5 6 Talo 85,4]-9 0 (Y = 26.73 + 0.56x, X = -29.38 + 11.43y)Riatse rareies Since the rey eS aia th gression line passes through the point (Z. 5). the mean valy Jota the FeRression equations can be used to find the value of constants, Ss of ana a=jy for regression equation of y on « eras iy ¥ -d¥ for regression equation of x on y re cal it cies seulated values ofa, b and ¢,d are substituted in the regression line y = dy respectively to determine the exact relationship. ed Example 14.1: 14.1: Use least squares regression line to estimate the increase in sales rey expect ted from an increase of 7.5 per cent in advertising expenditure, ee Firm Annual Percentage Increase Annual Percentage Increase in Advertising Expenditure in Sales Revenue Tommoower EDewosee ewnaenne Solution: Assume sales revenue (9) i n y) is dependent on advertising expenditu Calculations for regression line using following normal equations are shown in “Tble 14} Ey =na+bEx and Uxy = aEx + bEx? : Table 14.1: Calculation for Normal Equations Approach 1 (Normal Equations): Zy=na + bix or 40 = 8a + 56b Exy = ax + DEx’ or 378 = 56a + 524b Solving these equations, we get a= 0.072 and b = 0.704 Substituting these values in the regression equation y =a + bx = 0.072 + 0.704% For x = 7.5% or 0.075 increase in advertising expenditure, the estimated increase la sales revenue will be y= 0.072 + 0.704 (0.075) = 0.1248 or 12.48%Ox Oy oS 5, Example 14.3: The following data relate to the scores obtained by 9 salesmen ofa Compan, in an intelligence test and their weekly sales (in Rs 1000's) Salesmen : A B Cc D E F G H I Test scores : 50 60 50 60 80 50 80 40 70 Weekly sales: 30 60 40 50 60 30 70 50 60 (a) Obtain the regression equation of sales on intelligence test scores of the salesmen, (b) If the intelligence test score of a salesman in 65, what would be his expected weekly sales. [HP Univ. MCom, 1996] Solution: Assume weekly sales (y) as dependent variable and test scores (x) as independent variable. Calculations for the following regression equation are shown in Table 14.3. y-J= by (w- *®)= 50 1200 Substituting values in the re 120 os 1600 Dat eauation, we have ~50 y = 0.75 Ge — 60) TY = 5 + O.75e For test Score x = 65 of, salesmar in, we have 2 5 + 0.75 (65) = 55.75 ‘onclude that ih weekly sales is expected to be Rs 53.75 (in Rs 1000's) for a Points 5 I-T = dF) Table 144: Calculations for Reg@ ¥ Ww 9 ueger value, therefore deviations are taken from Since mean values ¥ and 7 are non- assumed mean as shown in Table 14.4. nEddy~ (Edy) Edy) _ 9x 65.40- 23.15 _ 'e = “ead 9x 484 —(2)" 58: 0.133 Substituting values in the regression equation, we have y~F =b,&-¥) or y~ 5.35 = 0.133 & - 15.22) = 3.326 + 0.133x (b)_ For job grade point x=20, the estimated average pay scale is given by 26 + 0.1330 = 3.326 + 0.133 (20) = 5.986 Hence, likely monthly pay for a job with grade points 20 is Rs 5986. Example 14.5: The following data give the ages and blood pressure of 10 women. Age : 56 42 36 47 49 42 60 72 63 55 Blood pressure 147 125 118 128 145 140 155 160 149 150 (@)_ Find the correlation coefficient between age and blood pressure. (b) Determine the least squares regression equation of blood pressure on age. (©) Estimate the blood pressure of a woman whose age is 45 years. [Ranchi Univ. MBA; South Gujarat Univ., MBA, 1997] Solution: Assume blood pressure (y) as the dependent variable and age (x) as the independent variable. Calculations for regression equation of blood pressure on age are shown in Table 14.5 Table 14.5: Calculations for Regression Equation Bes 455-49 dR Bloody 51 a ee 56 7 49 147 V3 4 14 42 -7 49 125 -20 400 140, 36 -18 169 18 -27 729 351 47 me 4 128 -17 289 34 @ea 0 0 Ges 0 0 0 42 =F. 49 140 -5 25 354 60 u 121 155 10 100110. 72 23 529 160 15 225 345 63 14 196 149 4 16 56 55 6 36 150 5 2.30 i $2. 1202 ia? 233 1813 itis (a) Coefficient of correlation between age and blood pressure is given by n3d,d, ~Zd,Edy ae ~ Gd, fnEd3 - Ed)? 10(1115) — (32) (-33) /10(1202) ~ (32)? /10(1813) - (- 33)? 11150+11 = 1h. 12206 _ 9 599 A /12020 1024, 18130-1089 13689 We may conclude that there is a high degree of positive correlation between age and blood pressure. (b) The regression equation of blood pressure dn age is given by y-F = be-¥) r=equation, we have 52.2) or y = 88.7584. 1y “sion equation of y on x DSe Ake is AB, the esti 58+1.11(45) eel average blood pressure will be = 89.758+49.95 = 138.708 ressure g P don re of a woman of 45 years is 134, General Sales Manager o “Kdymade men's wears toying on wn Enterprises—an enterprise dealing. ith the idea of increasing his sales to uring the last 10 years, it was found that 45,000, F = 30,000. Substituting these value ( ~ 30,000) 08 s (& ~ 45,000) = 0.5 ( - 45,000) ¥ = 30,000 + 0.5x ~ 2°,500 = 7500 + 0.5%. When a sales target is fixed at x = 80,000, the estimated amount likely to the spent on advertisement would be y = 7500 + 0.5% 80,000 = 7500 + 40,000 = Rs 47,500 Example 14.7: You are given the following information about advertising expenditure and sales: Advertisement (x) Sales () (Rs in lakh) (Rs in lakh) ye 7 ee eee ‘Arithmetic mean, = 10 WW Standard deviation, 6 3 wy Correlation coefficient = 0.8 ‘ Obtain the two regression equations. ; » Find the likely sales when advertisement budget is Rs 15 lakh. : ; Id be the advertisement budget if the company wants o attain sales target : wns 130 mkho {[Kumaon Univ., MBA, 2000, MBA, Delhi Unix, 2002] of Solution: (a) Regression equation of x on y is given by 6. a-B are » : : = = 12, F = 90. Substituting these values in the 8, 0, = 3,6, = 12 F ven ¥ = 10,7 see eee equation, we haveSMeSS STATISTICS yo 10= 08 3 9) or ew HO Regresion equation ofy on xf given by s oe ie cu * yom on ito) or ym a8 + 32x \ © 15 in regression equation of y on x. The likely average sales y= 584+ 5.2 (1D) = 58 + AK = 106 ulvertisement budget of Rs 15 lakh is Rs 106 lakh. on of x ony. The likely advertisement Thas the likely sales fo (6) Substiuating y = 120 in the regression eq bbudlget to attain desired sales target of Rs 120 lakh would be N= 8+ 0.2) =—8 + 0.2 (120) = 16 Hence, the likely advertisement budget of Rs 16 lakh should be sulficient to attain the sales target of Rs 120 lakh. Example 14.8: In a partially destroyed laboratory record of an analysis of regression data, the following results only are legible: Variance of x = 9 Regression equations : &x ~ 10y + 66 = 0 and 40x ~ 18y = 214 Find on the basis of the above information: (@) The mean values of x and y, (©) Coefficient of correlation between x and j, and (©) Standard deviation of y. [Pune Univ., MBA, 1996; CA May 1999) ) representing. Solution: (a) Since two regression lines always intersect at a point (%, imfad Yalues of the variables involved, solving given regression equations to get the mean. values ¥ and 7 as shown below: 8x — 10y ‘Multiplying the frst equation by 5 and subtracting from the second, we have S2y= 544 or y= 17, ie F = 17 Substituting the value of y in the first equation, we get 3, that is, F = 13 8x - 10(17) 66 or x= (6) To find correlation coefficient r between x and y, we need to determine the regression coefficients 6, and by. Rewriting the given regression equations in such away that the coefficient of dependent variable is less than one at least in one equation, 66 8 = 65 + 8 or y= aS Br ~ 10) = ~66 or 10y That is, b,, = 8/10 = 0.80 214 18 40x ~ 18y = 214 or 40x = 214 +1 = 418 By or fy or x = 4 That is, 6, = 18/40 = 0.45 Hence coefficient of correlation r between x and y is given by 1= JyXPq = J045X0.80 = 0.60 (©) To determine the standard deviation of y, consider the formula: 6, 6, 4 oe gw eh Cte, 0, ae ar 06 Example 14.9: There are two series of index numbers, P for price index and S for stock of a commodity. The mean and standard deviation of P are 100 and 8 and of S are 103 and 4 respectively. The correlation coefficient between the two series is 0.4. With theseoH Eaton Sime equation be used tatneetp’ (® Solution: The re ead off values oft Of values of for various values ofS. Can the an cause aie 8 for telus talento "ea off values of P for various values in given by Pea +S or (By ay Shy) B = 100, 5 = 19; tion, we have 1S Sp = 8:0, = 4,7 = O44, Substituting these values in the P~100= 948 478-103) or P= 17.6 + 0.85 equation cannot be Hes of § for various ae Seip Ull values of § for various values of P Thus to read OFF we use another regression equation of the form S=c+dP or s-3 = Leh Substituting given values in this equation, we b S-103 = 04 4 S-103 = 0.4 4 (100) or $= 83 + 02P Example 14.10: The wo re observations arey_ hm *® FeBression lines obtained in a correlation analysis of 60 Sx = 6x +24 and 10003 = 768: What is the co Wy = 768x ~ 9708 ofae cumues relation coefficient and what i its probable error? Show that the ratio. ae of variability of x to that of y is 5/24. What is the ratio of variances of x Solution: Rewriting the regression equations Se Gy +24 or x= S524 That is, by = 6/5 Bs 1060) = 76869708 or y= 7%, S708 ‘That is, by = 768/1000 10004 We know that}, = 7S = & and, = r2t = 788 therefore Lg, Tiga ro, | TOOK 6 768 bb, = P= 2x = 0.9216 ae 5 106 e2 Hence r = 0.9216 = 0.96. Since both 8, and 8,, are positive, the correlation coefficient is positive and hence 0.96. Probable error of r = 0.6745 4 0.6745 1=(0.96" PN, Yeo 0.0528 = =0. 77459_ Solving the given regression equations for x and y, we get ¥ = 6 and F = 1 because regression lines passed through the point (¥, ¥). o, 2 § ee ee ae "s, rae a o, - 6 ee 6, 5x0.96 4 5 Also the ratio of the coefficient of variability ot ear