ConductLogisticRegressionAnalysis

Introduction

When the outcome you're predicting is a category rather than a number — will this customer churn, which of three products will they pick — linear regression's continuous output doesn't map cleanly onto the question. ConductLogisticRegressionAnalysis fits the right model automatically: binomial logistic regression for a two-category outcome, or multinomial logistic regression when there are three or more, using statsmodels under the hood. It returns coefficients, p-values, and pseudo R-squared, and can plot probability curves per predictor so you can see how the predicted probability actually shifts across a variable's range.

Teaching Note

Logistic regression analysis is essential for:

Predicting binary outcomes such as customer churn, conversion, or default
Modeling multi-class choices like product preference or tier selection
Identifying key drivers that increase or decrease the probability of an event
Clinical research for disease diagnosis and risk factor assessment
Analyzing social science survey data with nominal or ordinal categories
Evaluating the odds ratios of specific predictors in classification tasks
Validating classification models through diagnostic visualizations

Parameters

Parameter	Default	Description
`dataframe`required	—	The pandas DataFrame containing the data for the regression analysis.
`outcome_variable`required	—	Name of the column in the DataFrame representing the categorical outcome variable.
`list_of_predictors`required	—	A list of column names for the independent (predictor) variables to include in the model.
`add_constant`	`True`	If True, adds an intercept (constant term) to the model. Defaults to True.
`scale_predictors`	`False`	If True, scales the predictor variables to have a mean of 0 and standard deviation of 1 using StandardScaler before fitting the model. Defaults to False.
`show_diagnostic_plots_for_each_predictor`	`True`	If True, displays logistic regression plots with probability curves for each predictor. Defaults to True.
`show_help`	`False`	If True, prints a quick guide to the console explaining how to access and interpret the output dictionary. Defaults to False.

Returns

A dictionary with 'Fitted Model' (the statsmodels Logit or MNLogit results object) and 'Model Summary' (a comprehensive statistical summary of the regression).

Example

python

from analysistoolbox.hypothesis_testing import ConductLogisticRegressionAnalysis
import pandas as pd

# Binomial logistic regression: predicting customer churn
df = pd.DataFrame({
    'churn': [0, 1, 0, 1, 0, 1] * 20,
    'monthly_spend': [50, 100, 45, 120, 60, 150] * 20,
    'tenure_months': [24, 2, 36, 1, 12, 4] * 20
})
results = ConductLogisticRegressionAnalysis(
    dataframe=df,
    outcome_variable='churn',
    list_of_predictors=['monthly_spend', 'tenure_months']
)
print(results['Model Summary'])