Logistic Regression: Sigmoid Decision Boundary

Logistic regression is the workhorse of binary classification: a linear score run through a sigmoid to turn it into a probability. This simulator renders that idea as a real 3D surface — a scatter of labeled points sits on the floor, and a rippled probability sheet floats above it, its tilt set by two feature weights and a bias. Drag the weight sliders to reshape the sheet and watch the decision boundary line slide across the floor, or click Train step to run one real batch gradient-descent update against the binary cross-entropy loss and watch the model converge on its own — live accuracy, loss and step-count readouts track exactly what a real training loop optimizes.