The Machine Learning “Advent Calendar” Day 11: Linear Regression in Excel

Linear Regression is introduced as a foundational weight-based model in machine learning, distinct from distance or rule-based methods. The goal is to find optimal parameters, a slope and an intercept, by minimizing a cost function like the Mean Squared Error (MSE). Two methods for finding these parameters are explained: the exact closed-form algebraic solution and the iterative optimization process of gradient descent. These core concepts of loss functions and gradient-based optimization are fundamental building blocks for more advanced models like neural networks.