Sie sind auf Seite 1von 4

ASSIGNMENT

REGRESSION TESTING
1.0 Consider a real or hypothetical problem involving one dependent variable and at least three
independent variables. (Problem definition).

Sol: Assuming there is a firm, Mojo Pizza. It wants to build a Sales forecasting model so that the firm can
generate monthly and region-wise forecasts using the following variables:

 Number of Delivery boys


 Advertising Spend
 Number of Outlets
 Pizza varieties served
 Competitor Activity (1= V Low; 7= V High)
 Customer base

Sale being the dependent variable here.

2.0 Data Set

There are 20 samples for 6 variables. Competitor Activity is a categorical variable while rest all are continuous
variable.

3.0 Analysis:

After Running the regression analysis with Sales as the dependent variables and Number of delivery boys,
Advertising speed, Number of Outlets, Pizza varieties served, Competitor activity, Customer base as
Independent variables we get the following matrix for the descriptive statistics.

Mean Standard Error Standard Deviation Count


Sales 25 4.802959614 21.47948838 20
Delivery_Boys 6.15 1.011187421 4.522167622 20
Advertising_Spend 11.15 1.054501628 4.715874647 20
Outlets 15 1.843908891 8.246211251 20
Pizza_Variety 13.3 0.44188591 1.976173868 20
Competitor_Activity 3.2 0.505756338 2.261811105 20
Customer_base 29.5 3.394655242 15.18135976 20

Table 1: Descriptive Statistics

3.1 Interpretation of R^2

Regression Statistics
Multiple R 0.975759909
R Square 0.952107399
Adjusted R Square 0.930003122
Standard Error 5.682811711
Observations 20
Assignment Multiple Regression QTDM
Table 2: R square

Interpretation: R^2 is. 952.Taken as a set, the predictor variables account for 95.2 percent of the variance in
the sales.

3.2 Interpretation of F statistic from the ANOVA table.

Let’s consider that confidence level is 95% and corresponding to that the significance level is 0.05

F=43.073, p=7.73^-7, p<0.001, R^2=0.952

Checking the significance of the regression:

Let a, b, c, d, e and f be the coefficient of number of delivery boys, advertising spend, number of Outlets,
pizza varieties served, competitor activity and customer base respectively.

Ho: All the coefficients are equal.

H1: at least one coefficient is unequal

The value of F at alpha 0.05 =2.92 which is less than 43.073, so we reject the null hypothesis and accept the
alternate hypothesis that at least one variable is unequal and we procced further

We now need to compare the significance value with the p value obtained in the anova table to check the
significance of the test (R square is significantly greater than 0).

As per the table, the value of R ^2 is greater than the p value (p value is coming to be 0.0000092) which
means that the predictors amount for the variance in sales. The regression model is significant.

ANOVA
df SS MS F Significance F
Regression 6 8346.173 1391.029 43.07344647 7.73371E-08
Residual 13 419.8265 32.29435
Total 19 8766

Table 3: Anova Table

Anova Table (test using alpha =0.05)

The overall regression model was significant.

3.3 The significance of regression coefficients and the regression equation.

Assignment Multiple Regression QTDM


Coefficients Standard Error t Stat P-value
Intercept 1.900283801 12.41652226 0.153044771 0.880713
X Variable 1 1.360861448 0.726927033 1.872074342 0.083862
X Variable 2 0.036211222 0.889072262 0.040729222 0.968131
X Variable 3 1.458032078 0.431551245 3.378583879 0.004942
X Variable 4 -0.862550045 1.001361338 -0.861377419 0.404639
X Variable 5 -1.15075931 0.792900775 -1.45132827 0.17039
X Variable 6 0.257983979 0.193571662 1.33275696 0.2055
Table 4: Regression Coefficient Table

The value of constant is 1.900.

Let’s see the significance value of the independent variables and we will be evaluating it with an
alpha of 0.05.

Predictor Significance /Non- P value


Significant

Delivery boys Non-Significant 0.083 (more than 0.05)

Advertising Spend Non-Significant 0.96 (more than 0.05)

Outlets Significant 0.004 (less than 0.05)

Pizza Variety Non-Significant 0.40 (more than 0.05)

Competitor Activity Non-Significant 0.17 (more than 0.05)

Customer Base Non-Significant 0.20(more than 0.05)

Table 5: Sales and individual variable significance

The regression equation is:

Sales= 1.9 +1.36*Delivery Boys + 0.036*Advertisement spend+1.45*outlet’s -0.86*Pizza variety


served -1.15* competitor activity +0.25*customer base.

Predictor Coefficient Relation with Sales


Delivery boys Positive Directly proportional
Advertising Spend Positive Directly proportional
Outlets Positive Directly proportional
Pizza Variety Negative Inversely proportional
Competitor Activity Negative Inversely proportional
Customer Base Positive Directly proportional

Assignment Multiple Regression QTDM

Das könnte Ihnen auch gefallen