Lab 6 Grading Guide
Model Selection Order
Based on my code, the model selection process should go:
cls_credits \(\rightarrow\) bty_avg \(\rightarrow\) gender
In Step 4 they are given two variables (cls_perc_eval, eval_completion) which both increase adjusted \(R^2\) by 0.018, not the 0.02 which is required to add another variable. So, everyone should stop at Step 4.
Question 1
To earn a Success:
- fits other 7 models (rank, pic_outfit, pic_color, large_class, eval_completion, cls_level)
- finds other 7 adjusted \(R^2\) values
- states top model includes cls_credits variable
If they do not state the top model or what variable was chosen:
Careful! Technically, you told me the name of the object that contains the model you decided was best. I want you to tell me the name of the variable(s) included in that model!
Question 2
To earn a Success:
- fits 13 models that include the variable chosen in #1 as an explanatory variable
- finds 13 adjusted \(R^2\) values
- states top model includes cls_credits & bty_avg
The response needs to indicate in some way that there are two variables included in the model. This can be by saying bty_avg was “added” to the previous model or by stating both variables included in the model.
If they do not indicate that there are two variables in the model (e.g., “beauty average was chosen” or “beauty score was included”):
Careful! Your model contains more variables than just bty_avg. Be specific about naming every variable included in your top model!
Question 3
To earn a Success:
- modifies code to fit models that include the variables chosen in #1 and #2 as an explanatory variables
- states top model includes all the variables they’ve chosen (cls_credits, bty_avg, gender)
If they don’t include a variable from #1 or #2 as an explanatory variable
Careful! Thus far, you’ve chosen two variables to be included in your model. So, your code (the map() function and select() function) need to have both these variables included, not just one!
The response needs to indicate in some way that there are two variables included in the model. This can be by saying bty_avg was “added” to the previous model or by stating both variables included in the model.
If they do not indicate that there are three variables in the model (e.g., “gender was included in the model” or “gender was chosen”):
Careful! You were specifically asked to name EVERY variable included in your top model.
Question 4
To earn a Success:
- modifies code to fit models that include the variables chosen in #1, #2, and #3 as an explanatory variables
- state the top model is the model from Step 3 (cls_credits, bty_avg, gender)
If they don’t include a variable from #1, #2, or #3 as an explanatory variable
Careful! Thus far, you’ve chosen three variables to be included in your model. So, your code (the map() function and select() function) need to have both these variables included, not just one!
If they choose to add either cls_perc_eval or eval_completion:
Careful! Look back at the adjusted R-squareds for these variables. Did they increase 2% (0.02) from your last adjusted R-squared?
Question 6
To earn a Success:
- states that the
step()function did not choose the same model - state that this difference was caused by different criteria (or the use of AIC)
If they don’t talk about the step() function using a different model comparison criteria:
Look back at the output from the step() function, specifically the first line of the output. Based on this line, do you think the step() function is also using adjusted R-squared to decide what model is best? If not, what criteria do you think the step() function is using?