Loss(Score) Function for Boosted Tree or Random Forest (JMPpro)

lucy_luo_conoco

If I understand correctly, Random Forest and Boosted tree model in JMPpro using R2 or RMSE as function to judge training/validtion model. If my data (no missing value) want to use minimize Absolute Percent Error min[(pred-true)/true] as function to train the model, how should I do it in JMPpro? or jsl coding? Thank you!

Byron_JMP

Unless there is something new in v14 that I haven't discvered yet, the loss/training funciton is not an option.

However, I highly recommend that you send a note to support@jmp.com and describe the tuning functions you would like to have as options.