Pandas for Everyone: Python Data Analysis, First Edition

13. Generalized Linear Models

13.1 Introduction

Not every response variable will be continuous, so a linear regression will not be the correct model in every circumstance. Some outcomes may contain binary data (e.g., sick, not-sick), or even count data (e.g., how many heads will I get). A general class of models called “generalized linear models” (GLM) can account for these types of data, yet still use a linear combination of predictors.

13.2 Logistic Regression

When you have a binary response variable, logistic regression is often used to model the data. Here’s some data from the American Community Survey (ACS) for New York.

Click here to view code image

import pandas as pd acs = pd.read_csv('../data/acs_ny.csv') print(acs.columns)

Index(['Acres', ...