Weighted Least Squares Estimation: IRLS Algorithm

This lecture covers the Iteratively Reweighted Least Squares (IRLS) algorithm for weighted least squares estimation. The instructor explains the heuristics behind the sampling distribution of the Maximum Likelihood Estimator (MLE) in Generalized Linear Models (GLM). The lecture delves into the iterative process, the score expression, and the importance of the weight matrix in the estimation process. Additionally, the lecture discusses the asymptotic normality of MLE in GLM, goodness of fit measures, nested models comparison using deviance, and diagnostic tools like leverage and Cook's statistic.