Best Model

Solution Preview

I use the econometric method to build a sales model given the data provided. I explore the individual variables, their relationships to one another, then provide results from a few different specifications. Theory asside it seems that time and number of employees are the best predictors of how many snowmobiles this firm will sell.

I usually start a project like this by looking at the data in a matrix. First, look at the last row (see the blue arrow). This row gives us an idea about the relationship between our dependent variable "Sales" and our independent variables. First we see a strong relationship between sales and date. Nice and linear. As time goes by sales steadily increase although there is a dip in 1993. Some sort of shock. I would want to include this trend variable in the model. Next, sales and expenses seem to be positively correlated too. Sales and number of employees is not as obvious, we'll explore this more in a second. There does appear to be a positive relationship between sales and rainfall. But it is tenuous.

When constructing a model we have two concerns the data and the theory. Regarding theory we have to ask "what causes snowmobile sales?" I would think a time trend variable would be good to include. The expenses that the company faces is certainly correlated to sales because as you sell more your costs are higher. However I don't know if I would say higher expenses for a firm leads to higher sales. It is more likely that high sales lead to higher expenses. If those are in fact firm expenses then I don't know that I would include it unless you can ...