Shelf Location Residual Plot

2500

2000

1500

1000
Residuals

500

-500

-1000

-1500

-2000

-2500
0 0 0 1 1 1 1
Shelf Location
Durbin-Watson Calculations

Sum of Squared Difference of Residuals 32516691.98

Sum of Squared Residuals 44232935.2

Durbin-Watson Statistic 0.735123994

 Dispensers Residual Plot
2500

2000

1500

1000
Residuals

500

-500

-1000

-1500

-2000

-2500
0 0 0 1 1 1 1
Dispensers
Regression Analysis
Coefficients of Partial Determination

Intermediate Calculations
SSR(X1,X2) 7860742.244
SST 52093677.44
SSR(X2) 624790.6176 SSR(X1 | X2) 7235951.626
SSR(X1) 6961300.087 SSR(X2 | X1) 899442.1568

Coefficients
r2 Y1.2 0.14058885
r2 Y2.1 0.019928978
Regression Analysis
All but Dispensers
Regression Statistics
Multiple R 0.36555495
R Square 0.133630422
Adjusted R Square 0.106556372
Standard Error 1187.597066
Observations 34

ANOVA
df SS MS F Significance F
Regression 1 6961300.08679 6961300.087 4.9357382845 0.03350275089
Residual 32 45132377.3544 1410386.792
Total 33 52093677.4412

Coefficients Standard Error t Stat P-value Lower 95%

Intercept 2696.631579 272.45345255 9.897586372 2.914709E-11 2141.66205937
Shelf Location 911.3017544 410.191016432 2.221652152 0.0335027509 75.7699995413
Upper 95%
3251.601099
1746.833509
Regression Analysis
All but Shelf Location
Regression Statistics
Multiple R 0.1095152859
R Square 0.0119935978
Adjusted R Square -0.018881602
Standard Error 1268.2281787
Observations 34

ANOVA
df SS MS F Significance F
Regression 1 624790.617647 624790.61765 0.3884540933 0.53753040157
Residual 32 51468886.8235 1608402.7132
Total 33 52093677.4412

Coefficients Standard Error t Stat P-value Lower 95%

Intercept 3234.2352941 307.590514008 10.514743293 6.546234E-12 2607.69392279
Dispensers -271.1176471 434.998676568 -0.623260855 0.5375304016 -1157.1809518
Upper 95%
3860.776665
614.9456577
Confidence and Prediction Estimate Intervals

Data
Confidence Level 95%
1
PHStat User Note:
Shelf Location given value 0
Dispensers given value 0 Enter the values for the given X's
in the cell range B6:B7.
X'X 34 15 17 (You can interactively change
15 15 8 these values at any time.)
17 8 17
To delete this note:
Select this note and then
Inverse of X'X 0.079122 -0.049296 -0.055924 select Edit | Cut.
-0.049296 0.119718 -0.007042
-0.055924 -0.007042 0.118061

X'G times Inverse of X'X 0.079122 -0.049296 -0.055924

[X'G times Inverse of X'X] times XG 0.079122

t Statistic 2.039513
Predicted Y (YHat) 2850.99

For Average Predicted Y (Yhat)

Interval Half Width 685.2778
Confidence Interval Lower Limit 2165.712
Confidence Interval Upper Limit 3536.267

For Individual Response Y

Interval Half Width 2530.777
Prediction Interval Lower Limit 320.2129
Prediction Interval Upper Limit 5381.766
te:

for the given X's

B6:B7.
tively change
any time.)
ote:
and then
.
Regression Analysis
Regression equation
Regression Statistics Bars sold = 2850.98964374482 + (930.739436619719
Multiple R 0.38845370643
R Square 0.15089628204 0.15089628204 15.08% of variation in bar sales is explained by variation in shelf locatio
Adjusted R Square 0.09611539701 9.61% of variation in bar sales is explained by variation in shelf location & dispensers,ta
Standard Error 1194.51616872
Observations 34

ANOVA
df SS MS F Significance F
Regression 2 7860742.24358 3930371.1218 2.7545426102 0.07923033809
Residual 31 44232935.1976 1426868.8773
Total 33 52093677.4412

Coefficients Standard Error t Stat P-value Lower 95%

Intercept 2850.99 336.000623577 8.4850724781 1.387939E-09 2165.71185664
Shelf Location 930.74 413.30658159 2.2519347092 0.031554018 87.7951092696
Dispensers -325.87 410.436362271 -0.793952621 0.4332588855 -1162.9575022

RESIDUAL OUTPUT

Observation Predicted Bars Residuals

1 3455.86205468 685.137945319
2 3455.86205468 386.137945319 Residuals wrt Time
3 2850.98964374 205.010356255
2500
4 2525.12261806 993.877381939
5 3781.72908036 444.270919635 2000
6 2525.12261806 2104.87738194 1500
7 2525.12261806 981.877381939
1000
8 3455.86205468 298.137945319
9 3781.72908036 1218.27091964 500
10 3781.72908036 1338.27091964 0
11 2525.12261806 1485.87738194 0 5 10 15 20 25 30
-500
12 2850.98964374 2164.01035626
13 2850.98964374 -934.98964374 -1000
14 2850.98964374 -2175.9896437 -1500
15 3781.72908036 -145.72908036
-2000
16 3455.86205468 -231.86205468
17 2850.98964374 -555.98964374 -2500
18 2525.12261806 204.877381939
19 2850.98964374 -232.98964374
20 3455.86205468 965.137945319
21 3455.86205468 657.137945319
22 2850.98964374 895.010356255
23 2850.98964374 681.010356255
24 2850.98964374 974.010356255
25 2525.12261806 -1429.1226181
26 2525.12261806 -1764.1226181
27 3455.86205468 -1367.8620547
28 2525.12261806 -1705.1226181
29 3455.86205468 -1341.8620547
30 2850.98964374 -968.98964374
31 3781.72908036 -1622.7290804
32 2525.12261806 -923.12261806
33 3781.72908036 -427.72908036
34 3781.72908036 -854.72908036
964374482 + (930.739436619719 X Shelf Location) + (-325.867025683513 X Dispensers)

xplained by variation in shelf location & dispensers

on in shelf location & dispensers,taking into account the sample size and number of independent variable

Testing the Assumptions

Since there are no patterns in the plots against independent variables, the model is appropriate
Ref to E10:
Dl=1.33
Du=1.58 Since D is less than Dl, these is +ve autocorrelation among the residual. Hence least squ
Durbin Watson statistic: 0.735124

F test- overall significance of the model

Upper 95% ref to tab. E5, F=3.32
3536.267431 Since F statistics is greater than 3.32, do not reject Ho.
1773.683764 Thus shelf location & dispensers are not related to the sales of the bars
511.2234508 Hence the explanantory variables do not explain the dependent variable

T test - Individual Significance test

Since P value for 'dispensers' is >.05, hence dispensers don’t affect Bar sales
But since P value for shelf location is <.05, shelf location affects bar sales

Coeff of partial Determination

0.140589 Proportion of variation in bar sales explained by shelf location
iduals wrt Time 0.019929 Proportion of variation in bar sales explained by dispenser

ANSWERS
1 The Claim isnt consistent: Increase in coupon dispenser decreases sales od b
Rest all explained above

2 Among the 2 I would recommend using only shelf location to increase bar sale

3 Price and promotion can also be used to determine the bar sales in addition to
15 20 25 30 35 40
es, the model is appropriate

ong the residual. Hence least square method is not appropriate.

dispensers don’t affect Bar sales

helf location affects bar sales

s explained by shelf location

s explained by dispenser

n dispenser decreases sales od bars

shelf location to increase bar sales

ermine the bar sales in addition to shelf location. All of em have have gud enuf individual significance.
Price Promotion Bars Shelf Location Dispensers
59 200 4141 1 1
59 200 3842 1 1
59 200 3056 0 0
59 200 3519 0 1
59 400 4226 1 0
59 400 4630 0 1
59 400 3507 0 1
59 400 3754 1 1
59 600 5000 1 0
59 600 5120 1 0
59 600 4011 0 1
59 600 5015 0 0
79 200 1916 0 0
79 200 675 0 0
79 200 3636 1 0
79 200 3224 1 1
79 400 2295 0 0
79 400 2730 0 1
79 400 2618 0 0
79 400 4421 1 1
79 600 4113 1 1
79 600 3746 0 0
79 600 3532 0 0
79 600 3825 0 0
99 200 1096 0 1
99 200 761 0 1
99 200 2088 1 1
99 200 820 0 1
99 400 2114 1 1
99 400 1882 0 0
99 400 2159 1 0
99 400 1602 0 1
99 600 3354 1 0
99 600 2927 1 0
 Index
End Aisle 1
Normal 0

Yes 1
No 0