Hands-On Data Analysis with Pandas

Interaction terms and polynomial features

We discussed the use of dummy variables back in the Preprocessing data section of Chapter 9, Getting Started with Machine Learning in Python; however, we merely considered the effect of that variable on its own. In our model that tries to predict red wine quality by chemical properties, we are considering each property separately. However, it is important to consider if the interaction between these properties has an effect. Perhaps when the levels of citric acid and fixed acidity are both high or both low, the wine quality is different than if one is high and one is low. In order to capture the effect of this, we need to add an interaction term, which will be the product of the features.

We may ...