This preview has intentionally blurred sections. Sign up to view the full version.View Full Document
Unformatted text preview: Handout 14 Model selection: all possible regression models Electric Bill data: Y =annual electricity bill, X 1 =monthly income, X 2 = number of persons, X 3 = area. . Largest model under consideration: Y = β + β 1 X 1 + β 2 X 2 + β 3 X 3 + ε , k = 3. Models to be considered: i) no predictor variable: Y = β + ε ii) regression with one predictor variable: Y = β + β 1 X 1 + ε , Y = β + β 2 X 2 + ε , Y = β + β 3 X 3 + ε, iii) regression with two predictor variables: Y = β + β 1 X 1 + β 2 X 2 + ε , Y = β + β 1 X 1 + β 3 X 3 + ε , Y = β + β 2 X 2 + β 3 X 3 + ε . iv) regression with three predictor variables: Y = β + β 1 X 1 + β 2 X 2 + β 3 X 3 + ε. The goal is to select the most appropriate of these 2 3 = 8 models. For any model with p- 1 predictor variables, the number of beta parameters estimated is equal to p and we will denote the residual sum of squares by SSE p . Note that df ( SSE p ) = n- p, MSE p = SSE p / ( n- p ) , R 2 p = SSR p /SSTO = 1- SSE p /SSTO,...
View Full Document
This note was uploaded on 09/28/2011 for the course STA STA 108 taught by Professor Jiang during the Summer '09 term at UC Davis.
- Summer '09