EN Business Decision Classifier — Logistic Regression (2D)
A Canvas2D logistic regression lab: gradient descent bends a straight decision line to separate two outcome classes, weights and loss updating live, point by point.
This 2D companion runs the exact same logistic-regression mechanic as the 3D version — z = w1·x1 + w2·x2 + b through a sigmoid, trained by mini-batch gradient descent on binary cross-entropy loss — but draws it on a plain Canvas2D context instead of an orthographic Three.js scene. Pick a dataset preset, watch the shaded decision boundary rotate into place, and track the loss curve falling as accuracy climbs.
Canvas2D logistic regression: sigmoid activation, mini-batch gradient descent on cross-entropy loss, a live decision-boundary heatmap and a loss-vs-step chart, all redrawn every frame.
2D · HTML5 Canvas 2D · 60 FPS target · runs fully client-side, no install