Summary

In this chapter, we used bivariate and multivariate statistical techniques and visualizations to get a better sense of bivariate relationships among features. We looked at common statistics, such as the Pearson correlation. We also examined bivariate relationships through visualizations, with scatter plots when both features are continuous, and with grouped boxplots when one feature is categorical. The last three sections of this chapter explored multivariate techniques for examining relationships and identifying outliers, including machine learning algorithms such as KNN and Isolation Forest.

Now that we have a good sense of the distribution of our data, we are ready to start engineering our features, including imputing missing values and encoding, transforming, and scaling our variables. This will be our focus for the next two chapters.