Q1:
Monthly sales (INR) = 32249.334 + 311.0897*Store Size (Sq. Ft.)
Increase in monthly sales is associated with approximately 311 per sq. unit of store size, given all other variables are constant.
The linear relationship between the monthly sales and store size is positive since both the variables increase in tandem with
each other.
Q2. Based on the evidence presented in the dataset, it can be
inferred that increase in the monthly sales is associated with
the increase in the store size (sq ft).
Q3. Using hypothesis test,
Ho : Sales per sq ft <= Rs. 500 is not applicable.
Ha : Sales per sq ft > Rs. 500 is applicable.
Based on the hypothesis test conducted we can conclude that the we do not have enough evidence to support the Null
hypothesis and hence we reject the same. Therefore, we can conclude that retail industry wisdom of Rs. 500 per month per sq.
ft. is applicable to CCD
Q4:
Prediction Interval for 200 & 500
200 Sqft = 83196.949896 105737.59707
500 Sqft = 164920.73763 210667.62639
We are 95% confident that the sales performance for Store with an area sq.ft. of 200 will be between
83196.94989 to 105737.59707 & for store with an area sq.ft of 500 will be between Rs. 164920.73763 to Rs. 210667.62639
Q5:
As we can observe in the graphs the data points in each graph appear to be weakly co-related and we can observe
there are a good number of Outliers in the plot. These do not lie on a straight line and hence we can confirm that
these plots are not reasonable for a regression model. We can also see that the Rsq value is 21.65% which is low and
hence conclude that the fit line is not the best-fitting line since the regression line would otherwise minimize the
difference between the actual and estimated results.
Q6:
Interpretation of Regression output:
1. The value of R-square represents fitment of good, so higher the R-square better the fit is;under given scenario. In this
case the Rsq value is 21.65% which is low and hence conclude that the fit line is not the best-fitting line.
2. If we compare R2 and R adj the values are only 21% approximately from which we can infer that the additional
variables are not adding any value to the model.
3. The P value in this case P < 0.0001
4. Multiple regression model equation = 0.0048217+ Whole market return *1.31568699+ IBM*0.22752
Q7: The estimated model appears to meet the conditions for the use of the Multiple Regression Model as we can see
the residual normal quantile plot the data points are almost normally distributed. Based on the residual by Predicted
plot we can see that there is no specific pattern observed and hence we can conclude that the graph is homoscedastic.
Q8.
Term Estimate Std Error t Ratio Prob>|t|
Intercept 0.0049275 0.007911 0.62 0.5338
Whole Market Return 1.5371567 0.175106 8.78 <.0001*
Simple regression model equation = 0.0049275 + Whole Market return * 1.537, slope = 1.537
Term Estimate Std Error t Ratio Prob>|t|
Intercept 0.0048217 0.007868 0.61 0.5405
Whole Market Return 1.3168699 0.204538 6.44 <.0001*
IBM Return 0.2275259 0.110768 2.05 0.0408*
Multiple regression model equation = 0.0048217 + Whole market retun * 1.3168 , Slope = 1.3168
When we compare the slopes of multiple regression model to that of simple regression, we can observe that the slopes are
different.
