Logistic Regression

Despite its name, Logistic Regression is a classification algorithm, not regression. It's the foundation of classification and still widely used today.

From Linear to Logistic

Linear regression outputs any number. But classification needs probabilities (0 to 1).

The Sigmoid Function

The sigmoid function squashes any number to the range (0, 1) - perfect for probabilities!

How Logistic Regression Works

Logistic regression = linear combination + sigmoid function

The Decision Boundary

Training with Gradient Descent

Practice: Logistic Regression Prediction

Key Takeaways

• Logistic regression is for classification despite its name
• Uses the sigmoid function to output probabilities (0 to 1)
• Creates a linear decision boundary
• Threshold (usually 0.5) converts probability to class label
• Uses gradient descent with cross-entropy loss for training
• In sklearn: LogisticRegression().fit(X, y).predict(X_new)

Next, we'll learn K-Nearest Neighbors - a completely different approach!