Lecture 2 Flashcards
Why can’t we use a linear model for probability values between 0 and 1?
What are the two options to have a linear model map to (0, 1)? Why are they equivalent?
What are three well-known examples of binomial regressions?›
Quickly explain the difference between a fixed design model and a random design model. When are each most likely used?
Why can a Poisson distributed random variable not be explained by a regualr linear model?
What are two examples of a Poisson regression?
Why can a Poisson regression not be rewritten into a linear regression model where the error term is Poisson distributed?
What is the idea of Generalized Linear Models?
What are the three elements of GLMs?
What are the two often used properties that follow from this element (of GLMs)?
Show that a Bernoulli Linear Model is a GLM?
What is the likelihood of a GLM? How is the MLE obtained?
What are the two main results of the GLM under regularity conditions?
What is different about computing the T-statistic for GLMs, than for a regular linear model?