Pandas for Everyone: Python Data Analysis, 2nd Edition

16

Model Diagnostics

Building models is a continuous art. As we start adding and removing variables from our models, we need a means to compare models with one another and a consistent way of measuring model performance. There are many ways we can compare models, and this chapter describes some of these methods.

16.1 Residuals

The residuals of a model compare what the model calculates and the actual values in the data. Let’s fit some models on a housing data set.

Click here to view code image

import pandas as pd
housing = pd.read_csv('data/housing_renamed.csv')

print(housing.head())

 neighborhood type units year_built sq_ft income \ 0 FINANCIAL R9-CONDOMINIUM 42 1920.0 36500 1332615 1 FINANCIAL R4-CONDOMINIUM 78 1985.0 126420 6633257 2 FINANCIAL ...