Feature Engineering Made Easy

Exploratory data analysis

Now we can conduct some exploratory data analysis. Since the purpose of polynomial features is to get a better sense of feature interaction in the original data, the best way to visualize this is through a correlation heatmap. 

We need to import a data visualization tool that will allow us to create a heatmap:

%matplotlib inline
import seaborn as sns

Matplotlib and Seaborn are popular data visualization tools. We can now visualize our correlation heatmap as follows:

sns.heatmap(pd.DataFrame(X_poly, columns=poly.get_feature_names()).corr())

.corr is a function we can call on our DataFrame that gives us a correlation matrix of our features. Let's take a look at our feature interactions:

The colors on the heatmap