Additional Topics In ANCOVA

1
Additional Topics In ANCOVA
2
Model Validation

• Several different methods for this
• We will divide the data into two groups in some
  type of random manner.
• The training group and the validation group -
  80/20 split
• Run Snells volume calculation. Volume 1

3
Model Validation Cont

• Do full analyses on training set. Get your best
  model. Fix all problems
• Do exploratory on validation set.
• Use model in part 1. and use all the data and use
  group as a factor.
• Test to see if parameters are equal across
  groups. Verify all assumptions.

4
Subset Selection In ANACOV

• Much more complex
• Uses - stepwise

5
Steps

• Setup Model
• Get Design Matrix
• Merge design with original data
• Fit Y versus factors save residuals - r
• Fit multivariate (factorscovariates) versus
  factors save residuals rr
• Fit r versus rr use stepwise
• Take final model Use ANCOVA