A Data Science Central Community
I am building random forest on employee churn data where %events(churns) is 8%.
Is there any upper cutoff on the number of trees I mention in the RF command?
Which accuracy metric should I use to compare one model to other-Sensitivity/specificity /TPR/TNR/Overall acuracy/AUC?
After scoring the validation data,I am sorting the prbabilities in descending order and creating quartiles.The count of events in quartiles is not following decreasing trend perfectly. Any ideas as to what might be the possible reasons?