Core Machine Learning Concepts

Explain the bias-variance tradeoff and its relationship to model complexity, overfitting, and underfitting.

Define and compare common loss functions for regression (MSE) and classification (cross-entropy).

Implement logistic regression for binary classification, explaining its connection to probabilistic modeling.

Derive and implement gradient descent to optimize the parameters of a logistic regression model.

Apply L1 and L2 regularization to a linear model to mitigate overfitting.

Select and justify appropriate evaluation metrics for classification tasks (accuracy, precision, recall, F1-score).

Implement k-fold cross-validation to obtain a robust estimate of model generalization performance.