What does logistic regression do?

Logistic regression models the probability of a binary outcome (1/0, yes/no) as a logistic function of one or more predictors. It is widely used in statistics and machine learning for classification problems.

What format should my data be in?

Paste rows where the last column is the binary target (0 or 1) and the previous columns are numeric features, separated by commas or spaces. Optionally include a header in the first row.

Can this replace scikit-learn or R?

This is a light, in-browser implementation for learning and quick checks. For production models, large datasets, regularization, and multi-class problems, use a full ML library.

Logistic Regression Calculator

Train a binary logistic regression model directly in your browser. Paste your data, click train, and get the intercept, weights, predicted probabilities, and log-odds interpretation.

1. Paste your dataset

Format: each row = 1 observation. Last column = target (0 or 1). Columns separated by comma, semicolon, tab, or space. First row can be a header.

2. Training parameters

Learning rate

Epochs

Decimals

3. Model output

No model trained yet.

4. Predict with the trained model

After training, we’ll show the feature list. Enter values in order and get the probability \( P(y=1 \mid x) \).

Logistic regression essentials

Logistic regression models the log-odds (logit) of the positive class as a linear combination of predictors:


                logit(p) = ln( p / (1 − p) ) = β₀ + β₁x₁ + β₂x₂ + … + βₖxₖ


                p = 1 / (1 + e^(−(β₀ + β₁x₁ + ... + βₖxₖ)))

During training, we minimize the logistic loss (cross-entropy) over all observations. Here we do it with batch gradient descent.

Good to know

Features on wildly different scales can slow or destabilize training. Normalizing helps.
If your target isn’t 0/1, map it first.
This page does a basic, educational implementation—no regularization, no auto class-weight.