Lecture 3 Flashcards
What is the meaning of a nested model?
When doing testing for significance between two nested models, what is the main idea of the test?
What is the test statistic of two nested models?
What is the distribtution of a test between nested models under the null hypothesis?
What is the idea of a classical linear model (compared to a linear model)?
How are residuals defined for GLMs?
What is the Log Likelihood Ratio test statistic for GLMs? What is its distribution?
What is the definition of the full model and the null model?›
i.e., the Null model has no explanatory variables (only a constant)
What does the following notation mean (in the context of prediction error)?
What does the following notation mean (in the context of prediction error)?
What is the definition of the squared prediction error?
To what can the SPE be rewritten that can be used to explain the error in prediction? What do all the three terms mean?
What is the variance-bias tradeoff?
How can we use the SPE to choose between models?