Efficiency of Decision Tree (C5.0)

classification
rmachinelearning

#1

I am using the decision tree (C5.0) for classification in R but i am getting the error more than 50% in training stage.

How can i improve the result?


#2

@hinduja1234 - could you please brief on how do you say that you are getting 50% error? Are you measuring the performance using the confusion matirx? C5.0 is one of the robust algorithms under the decision tree family. If you had chosen the right variables you should be getting good results.


#3

@karthe1-I have given the model name as m1 and use the summary command on model m1 during the training stage in which it tells about the error of the model.


#4

@hinduja1234 - kindly check whether the variables are actually contributing to the target variable. If that is correct, then you can also try boosting. Try increasing the number of trails.

In such situations, I normally try applying some other algoirthm, say for example random Forest, to see if still the issue persists.