An option to include R-squared with adj R-squared when choosing lambda_z_n

[cahn88]

My observations and simulations show that adj R-squared criteion chooses bigger lambda_z_n than R-squared. I hope you add an option to choose between R-squared and Adj R-squared as a criterion for lambda_z_n.

Thank you!

Jae

[billdenney]

Thanks for this suggestion. We can definitely implement something like this. To make the request more precise, here are the details that I see:

Instead of using adjusted r^2, use the regular (unadjusted) r^2. That would require:

1. Make it so that the existing PKNCA.option() of "adjusted.r.squared.factor" allows the value to be NA (that would be done here).
2. Generalize the function above so that it would work for both regular r-squared and adjusted r-squared (e.g. take in an argument of what type of r-squared it is and modify the messages).
3. Make a new PKNCA.option() called "r.squared.factor". It would validate the same way as "adjusted.r.squared.factor"
4. Make pk.calc.half.life() take another parameter r.squared.factor = NULL (just below this line).
5. Validate that one but not both of the r.squared.factor or adj.r.squared.factor was not NA within pk.calc.half.life().
6. Change these lines so that either adj.r.squared.factor or r.squared.factor was used.
7. Add testing for all the new steps above.