Lecture 10: Inference Concerning GLMs
(Text Sections 5.3-5.4) Recall that the Taylor Series expansion of a function f (x) about a value x close to x is 1 f (x) = f (x ) + (x x )f (x ) + (x x )2 f (x ) + . . . . 2 Therefore, for an estimate which is close
Lecture 11: Model Adequacy, Deviance
(Text Sections 5.5-5.7) Deviance is an important idea associated with a fitted GLM. It can be used to test the fit of the link function and linear predictor to the data, or to test the significance of a particular pred
Lecture 5: Newton's Method, the Exponential Family
(Text Sections 4.2, 3.2) When there is no closed form for the solution of d log L(; y) g() = 0 d we may use a numerical method to determine the MLE. The Newton-Raphson method is one commonly used choice.
Lecture 2: More on Linear Models
(Text Sections 2.4, 6.2.2, 6.4, 6.5) Types of Predictor Variables Predictor variables are either numeric or categorical. Numeric variables take on meaningful numeric quantities. They are further classied as continuous or d
Lecture 7: Generalized Linear Models
(Text Sections 3.1, 3.4, 3.5) GLMs form a class of models that allow us to describe a variety of types of response variables. GLMs have 3 components: 1. The distribution of Yi , which we assume is in the exponential fa