Suppose that we fit three regression models (OLS linear regression) with training mean squared error MSE_1, MSE_2, and MSE_3. We know that the first model 1 has first 7 predictors, the second model 2 has additional two more predictors, and the last model has all k predictors (K > 9). What is the relationship among MSE_1, MSE_2 and MSE_3?
Asked
Active
Viewed 35 times
0
-
Splendid! Now what do you know about how parameter counts influence OLS performance? – Dave Feb 02 '22 at 22:30
-
@Dave I think that's what I'm missing here. Could you please enlighten me? – Telecaster Feb 02 '22 at 22:53
-
If you add an extra variable to the model, a variable whose coefficient can be zero, how does the affect the model performance? Hint: Think about what it would mean for that new coefficient to be zero. – Dave Feb 03 '22 at 01:32