For binary data case
1
May pls clarify if features returned in importance order?
2
And if there contradiction between features
For example one feature shows that with high probability should be label yes and another show that should be label no
Then what will be
Let's say one feature F1 is contradicting with other many features then this F1 feature will be marked as low importance and will not be returned?
Since by removing this feature we improve probability of correct classification
But keeping this F1 feature we get bad performance?
I ask this question since I see that one feature with high Likelihood for label YES
Is removed from list of returned good features....
Thank you very much in advance for taking care...
For binary data case 1 May pls clarify if features returned in importance order? 2 And if there contradiction between features For example one feature shows that with high probability should be label yes and another show that should be label no Then what will be Let's say one feature F1 is contradicting with other many features then this F1 feature will be marked as low importance and will not be returned? Since by removing this feature we improve probability of correct classification But keeping this F1 feature we get bad performance?
I ask this question since I see that one feature with high Likelihood for label YES Is removed from list of returned good features.... Thank you very much in advance for taking care...