Logistic Regression Calculator

Modeling a Yes/No Outcome

Linear regression predicts a continuous number, but plenty of real questions have a binary answer: will this customer churn, did the patient recover, did the loan default. Logistic regression is built for exactly that case. Rather than fitting a straight line, it fits an S-shaped curve that stays bounded between 0 and 1, so its output can be read directly as a probability.

The Formula

p = 1 / (1 + e−(b0 + b1·x))

This calculator fits b0 (intercept) and b1 (slope) using batch gradient descent on the logistic log-loss, working on standardized x values internally for numerical stability, then converts the coefficients back to the original x scale.

Worked Example

x = 1, 2, 3, 4, 5, 6. y (binary outcome) = 0, 0, 0, 1, 1, 1 — a clean split where the outcome flips from 0 to 1 partway through the range.

Logistic regression fit result
QuantityValue
Intercept (b0)−22.9447
Slope (b1)6.5556
Predicted probability at x = 3.50.50

Computed by running the calculator's own gradient-descent fitting procedure (3,000 iterations, learning rate 0.3) on standardized x values.

The model places the 50% probability threshold almost exactly at x = 3.5, the midpoint between the last "0" (x = 3) and the first "1" (x = 4) — exactly where a clean binary split should put it.

Where This Calculation Matters

  • Churn and conversion modeling — predicting the probability a customer cancels a subscription or completes a purchase based on a single predictive measurement.
  • Medical risk scoring — estimating the probability of a binary clinical outcome (disease present or absent) from a single biomarker or risk factor.
  • Threshold-based classification — once fit, the model assigns a predicted class (0 or 1) by comparing the predicted probability against a 0.5 cutoff, or another threshold chosen for the application.
Note: This calculator fits a simple (single-predictor) logistic regression. Real-world classification problems often need multiple predictors, which requires a full statistical package rather than a single x variable.

How to Use This Calculator

  1. Enter the X values as a comma-separated list (at least 4 data points).
  2. Enter the Y values as a comma-separated list of 0s and 1s, matching each X value.
  3. Optionally enter a specific X value to predict its probability and predicted class.
  4. Select Calculate to get the fitted model and, if requested, the prediction.

Related Calculations

For predicting a continuous outcome instead of a binary one, use the Linear Regression Calculator. To measure how strongly the predictor relates to the outcome first, see the Pearson Correlation Calculator.