Business Decision Classifier — Logistic Regression (2D)

Dataset

Training

Learning rate η: 0.10

Stats

Step0

Loss (BCE)—

Accuracy—

w1 (engagement)—

w2 (tickets)—

bias b—

Red = predicted high churn risk (class 0), blue = predicted low risk (class 1). Opacity encodes confidence |p−0.5|×2. The line marks p = 0.5, the learned decision boundary.