Decision Tree using RapidMiner

tanthiamhuat

Can anyone explain to me in what circumstances do we use different criteria (gain_ratio, informaton_gain, gini_index, accuracy) for the Decision Tree?

MartinLiebig

Hi,

in the end those are different implementations. There is good reasons for them, but noone can tell you before hand which works best. This is somehow the difference between stats and data science. Simply try them out and measure performance in a X-Validation.

~Martin