"decision tree with weights"
Contributor I
Hello:
I'm working with a very skewed dataset, where target class distribution is 99.5% vs 0.5%.
I'm trying to build a decision tree. Regular decision tree operator assigns all the leaves to the majority class, and by X-Validation I trivially get 99.5% accuracy.
I added to examples a weight attribute, 1 or 200, correspondingly, as described onhere. However, "Decision Tree" operator ignores examples' weights by default and I could not find any way to enable it. I tried building "X-Validation" with enabled "use example weights option" check box in "Performance (classification)" operator, and it looks like examples' weights are actually used here, by "Performance (classification)" operator, (I get 50% accuracy instead of 99.5%, meaning that I assigned weights correctly).
Could you, please, tell me how I can build a decision tree which uses examples' weights?
I'm working with a very skewed dataset, where target class distribution is 99.5% vs 0.5%.
I'm trying to build a decision tree. Regular decision tree operator assigns all the leaves to the majority class, and by X-Validation I trivially get 99.5% accuracy.
I added to examples a weight attribute, 1 or 200, correspondingly, as described onhere. However, "Decision Tree" operator ignores examples' weights by default and I could not find any way to enable it. I tried building "X-Validation" with enabled "use example weights option" check box in "Performance (classification)" operator, and it looks like examples' weights are actually used here, by "Performance (classification)" operator, (I get 50% accuracy instead of 99.5%, meaning that I assigned weights correctly).
Could you, please, tell me how I can build a decision tree which uses examples' weights?
Tagged:
0
Answers
i want to use a decision tree , svm or nn but they all seem to ignore the weight column in my data set