Model Selection

1 Textbook Reading Part 1 – Due by Monday

Required Reading: Model Selection

Reading Guide – Due Monday by the beginning of class

Download the Word Document

2 Textbook Reading Part 2 – Due by Wednesday

Required Reading: Sampling Variability

Reading Guide – Due Wednesday by the beginning of class

Download the Word Document

3 Concept Quiz – Due Monday by the start of class

Question 1 – \(R^2\) is the [correlation / percent] of variation in the [explanatory / response] variable that is explained by the [response variables / explanatory variables] and the [decided model / best model / linear model].

Question 2 – Adjusted \(R^2\) should be used for [simple / multiple] linear regression and unadjusted \(R^2\) should be used for [simple / multiple] linear regression.

Question 3 – In the adjusted \(R^2\) formula \(n\) represents the [number of explanatory variables / sample size / response variable] and \(k\) represents the [number of explanatory variables / sample size / number of coefficients needed to be calculated].

Question 4 – When choosing a model based on adjusted \(R^2\), the model with the [lowest / highest] value should be chosen.

Question 5 – Backward selection starts with the model with [no / every] explanatory variable included.

Question 6 – Backward selection [adds / removes] variables from the model. If the [R-squared / correlation / adjusted R-squared] when the variable is [added / removed] is [smaller / larger], the variable will be [added / removed] from the final model.

Question 7 – Forward selection starts with the model with [no / every] explanatory variable included.

Question 8 – Forward selection [adds / removes] variables from the model. If the [R-squared / correlation / adjusted R-squared] when the variable is [added / removed] is [smaller / larger], the variable will be [added / removed] from the final model.

4 R Tutorials

Note that one is due on Monday and one is due on Wednesday!

Due Monday: Assessing Model Fit

Due Wednesday: Foundations of Inference