Generalized Linear Models: GLMs for Non-Gaussian Data

This lecture by the instructor covers Generalized Linear Models (GLMs) for non-Gaussian data, focusing on GLMs for regression with exponential family responses. The lecture delves into the interpretation of the natural link function, the asymptotic normality of the Maximum Likelihood Estimator (MLE) in GLMs, and measures of fit using deviance. It also discusses residuals, leverage, and the Cook statistic, providing insights into logistic regression for binary data and loglinear regression for count data. The lecture concludes with remarks on the scale parameter in GLMs.