Hands-On Data Analysis with Pandas

Analyzing residuals

Whenever we work with linear regression, we should visualize our residuals, or the discrepancies between the actual values and the model's predictions; as we learned in Chapter 7, Financial Analysis – Bitcoin and the Stock Market, they should be centered around zero and homoskedastic (similar variance throughout). We can use a kernel density estimate to assess whether the residuals are centered around zero and a scatter plot to see if they are homoskedastic.

Let's look at the utility function in ml_utils.regression, which will create these subplots for checking the residuals:

import matplotlib.pyplot as pltimport numpy as npdef plot_residuals(y_test, preds):    """    Plot residuals to evaluate regression.    Parameters: - y_test: ...