logistic regression

Logistic Regression

A classification model that predicts the probability of a binary outcome — churn vs retain, convert vs not — by modeling the log-odds of the event as a linear function of input variables.

When you'd see it: Churn prediction, fraud detection, credit scoring, conversion modeling, medical risk scoring. Despite the name, logistic regression is a classification model — it predicts probability (0 to 1) rather than an unbounded continuous number.

Why it matters: Logistic regression outputs a probability, which is more useful than a binary prediction because the business gets to decide the classification threshold. A credit card issuer can set different thresholds for different risk appetites. It's also interpretable — each coefficient tells you how much the log-odds changes per unit of that predictor.

Common mistakes: Conflating logistic regression with linear regression because they share the word 'regression.' The two models have different outputs (probability vs continuous number) and are evaluated differently (AUC/precision/recall vs R²/RMSE). They're related mathematically but used for completely different problems.

Related linear regression, AUC, confusion matrix, precision, recall, overfitting