Glossary

Logistic regression

A statistical model that estimates the probability of a binary outcome as a function of input variables.

Also called: logit model

Logistic regression estimates the probability that a binary outcome, such as "churned" versus "did not churn," equals one, as a function of one or more input variables. Despite the name, it is a classification method, not a continuous-outcome regression: it models a probability, then a threshold, commonly 0.5, converts that probability into a predicted class.

The model fits a linear combination of the inputs and passes it through the logistic function, p = 1 / (1 + e^-(b0 + b1x1 + ... + bnxn)), which squashes any real-valued input into a value between 0 and 1. Coefficients are typically fit by maximum likelihood rather than the least-squares method used in ordinary regression analysis, and each one describes the change in the log-odds of the outcome per unit change in its variable, which practitioners often convert to an odds ratio for interpretation.

Logistic regression remains widely used, alongside more complex models, because its coefficients are directly interpretable and it performs well as a fast, well-calibrated baseline. It assumes a roughly linear relationship between inputs and log-odds, and is evaluated with a confusion matrix and ROC AUC rather than the R-squared used for ordinary regression.

Last reviewed September 22, 2026

In the index now

Related terms

Classification A supervised learning task that assigns each input to one of a fixed set of discrete categories or labels. Regression analysis A statistical method for estimating the relationship between one outcome variable and one or more predictor variables. Confusion matrix A table comparing a classifier's predicted labels against actual labels, breaking results into true and false positives and negatives. ROC AUC (area under the ROC curve) A single number summarizing how well a classifier separates positive from negative cases across every possible decision threshold. Feature engineering Creating, transforming or selecting the input variables a model is trained on, to make the underlying pattern easier for it to learn.