Logistic Regression

Gradient descent logistic regression from scratch — log loss curve, confusion matrix, decision boundary (2-feature mode), accuracy, precision, recall, F1.

Presets

Hyperparameters

Features per row

α (learning rate)Max iterationsThreshold

Training Data

x₁x₂y

Did not converge in 1000 iterations — try larger α or more iterations

Log Loss Curve (1000 iterations)

Decision Boundary (x₁ vs x₂)

● class 1● class 0--- boundary

Model Coefficients

β₀ (bias)-5.988

β1 (x1)1.102

β2 (x2)1.255

Training

Final log loss0.06519

Iterations1000

Gradient norm0.02477

Classification (threshold 0.5)

Accuracy1

Precision1

Recall1

F11

TP3

FP0

FN0

TN3

Per-Row Predictions

#yP(1)ŷ

100.084970

200.12990

310.94721

410.94321

500.043920

610.99471

h(x) = σ(β₀ + β₁x₁ + … ) · J = -1/n Σ[y·log(h)+(1-y)·log(1-h)]