Extract features to explain different states of the world

Question

I have a problem that can be seen as the inverse of a classification problem. I don't need to classify points, but to explain the differences (if any) between points in different, pre-specified classes.

Let's say there are two states of the world: summer and non-summer. During summer, the sales of apples are down vs non-summer.

The hypothesis is that some people buy apples year-round and some stop or drastically reduce their apples consumption in summer. And/or some varieties of apples are way down in the summer whereas some others are relatively robust to the summer effect.

I have a dataset with apple features (varieties, etc) and their sales and buyers, and a dataset with features about the buyers.

What is a good technique or model that can tell me which features (about apples and/or buyers) do a good job of discriminating summer-sensitive apples/buyers from non summer-sensitive apples/buyers?

Answer 1

Use regression model with sales are dependent variable and features + season as explanatory variables.

In order to find out which features really matter, use stepwise regression. The final model will give you an idea which features are causing the sales to change.