Locating New Pam and Susan‘s Stores

...

A Scatterplot of Residuals against predicted sales is also obtained, and it is shown below:

[pic 6]

Note: Sales is given in thousands ($1000).

In addition, a Histogram of the Residuals obtained from the Regression output is shown below. As you can see, the Histogram is almost bell -shaped, meaning that the data is normally distributed.

[pic 7]

Questions

- As a possible alternative to the subjective “competitive type” classifications, how well can you forecast sales using the demographic variables ( along with the store size and the percentage of hard goods)? What does your model reveal about the nature of location sites that are likely to have higher sales?

In using the “competitive type” classifications in my model, the regression equation suggests that sales are high in locations with lower Competitive Types, high % of Spanish speaking population, low % of households with freezer, low percentage of households with dryer and with high population.

- How good is the “competitive type” classification method (along with using the store size and the percentage of hard goods) at predicting sales? What recommendation do you have for simplifying the competitive type categories?

The “competitive type” classification method is helpful in trying to evaluate what factors would affect sales and revneue for Pam and Sue stores. My recommendation is to simplfy the competitive factors as you would want to see what factors would have the most direct impact on sales. In looking at my final regression model, comtypes 1, 2 and 7 were highly significant with very small p-values. And in looking at my scatterplot as competition increases sales decreases.

- Two sites, A and B are currently under consideration for the next new store opening. Characteristics of the two sites are provided below in Table B which would you recommend? What forecasting approach would you recommend?

In order to decide which site would better for Pam and Susan, I would do a multiple regression model to help forecast the sales figures for both sites. I would recommend site A as it has a higher predicted sales than location B. The values of variables along with its contribution to the regression equation are given below. The estimated value of sales can be estimated from regression as follows:

[pic 8]

- Two variables in the data base are under managerial control: the size of the store (square feet of selling area) and the percentage of hard goods stocked in the store. Margins on hard goods (house wares, appliances, stationary, drugs) are different from margins on soft goods (clothing for example). What impact does these variables have on sales?

The Pearsons’ correlation coefficient between sales and various factors are given below. Clearly many factors have high correlation with the sales. The positive correlation implies that an increase in that variable is followed by an increase in sales. A negative correlation suggests that an increase in that variable is followed by a decrease in sales. High negative correlations are highlighted in yellow.

[pic 9]

- Technical: For your recommended model, check to make sure the technical assumptions are satisfied. Comment on any points that would concern you based on the diagnostics.

The important assumptions of regression analysis are normality and homogeneity of variance of residuals. The normality assumption is evaluated using the histogram of residuals.

[pic 10]

The histogram suggests that the residuals are approximately normally distributed.

[pic 11]

The plot of residuals against the predicted values are randomly distributed on either side of the line through zero. That is, the value of the residuals does not depend on the size of the predicted value. Thus we can conclude that , the homogenity of variance assumption is valid for the regression model.

Conclusion

In order to give a solid recommendation for the new Pam and Susan store location, we had to statistically compute several models that would encompass the economic and demographical data, population type, sales numbers, store size and the competitive types. The data is used to create our multiple regression model. Our model indicates that the main factors that would determine which location will produce higher sales are:

- Densely populated area with relatively little direct competition

- High income areas with little competition

- Store located along the side of the roads

- % spanishsp

- % dryers

- %freezer

- Population

It appears that convenience, size, income, hard goods, language and location would be the deciding factors for the new store location. A buyer wants to shop in an area that is familiar, convenient and within walking distance. With the rising price of gas, many people would like to reduce the amount of mileage in their car. Also, people with high income have money to spend in the stores. And reducing the competiveness would increase the potential for the store to have health sales figures. As Spanish is one of the most popular languages in the U.S, it would be important that they employees at Pam and Sue stores are bilingual. In addition, having hard goods such a dryers and freezers are important in helping to keep households clean and to help lower the amount of disposable income spend on food.

We can use our regression model to forecast future value of sales, and to illustrate the relationship between sales and the other variables. It was also able to compute the predicated sales in that area. Figuring out the best location between Site A and B is determined using our regression model, and the location that has the highest sales value will be the most ideal location to open a store. In this case, Site A has the highest forecasted value, and it is the most ideal location to open a new store.

...