I read that dividing the data into train, validation and test set to solve overfitting?
Or is there a better way for decision trees?
I read that dividing the data into train, validation and test set to solve overfitting?
Or is there a better way for decision trees?
Hello @dpeez
Its always a good thing to use a validation set to see if the outcome is about the same as for your test set. Be carefull when you use a decisiontree because they are very senstive to over- underfitting.
To reduce overfitting, starting your modelproces with a random forest model (an esemble of trees) is a beter idea.
For example take a look at at this RandomForest model from the KNIME hub for more information and inspiration
gr
Hans
Thankyou !!! any idea for this? How to find a target attribute in a test set using a decision tree or Random forest or SVM classifier built from the training set?
This topic was automatically closed 182 days after the last reply. New replies are no longer allowed.