@dlehman I tried two things. Clean up and calculate the Regression with the data you provided. It will result in some missing values in the prediction which would have to be removed and like has been mentioned doing it without proper splitting will do you no good.
I used some H2O.ai auto-machine-learning which via R also provides with a variable importance list. You might see for yourself which variables are worth exploring and which might be some sort of leak. The results are in a subfolder /model/validate/ of the workflow.
If you want to explore more tradition Regression models you might take a look here: