-1

I have four predictor variables to model a soil property. Each of these predictor variables is generated by ten different methods (four groups of ten). Is there an algorithm (in R, Python, and ...) that selects the best type from each of these groups to predict soil properties?

Thanks.

Javad
  • 11
  • 1
  • 2
    Please provide a reproducible example table of the variables, and probably an example of what you consider best. Otherwise it's really hard to know what you're after. Welcome to Stack Overflow by the way, sorry I can't be more helpful. – Sethzard Jun 20 '22 at 08:00
  • There are lots of ways to measure predictive power. Your problem sounds quite unusual though, so would probably involve some manual programming to test out the different combinations of predictors/methods - I can't think of a pre-built tool you could use. – rw2 Jun 20 '22 at 08:14

1 Answers1

0

Try to use PyImpetus, it is a Markov Blanket based method to find the best features. https://github.com/atif-hassan/PyImpetus. Rest, if you need specific help, kindly mention the features for anyone to reproduce the results on.

Hetarth Chopra
  • 136
  • 6