Pages: [1]
Author Topic: Comparing multiple methods with same dataset in the Wrapper Feature Selection  (Read 441 times)
Posts: 6

« on: June 27, 2014, 04:27:08 PM »

As part of a project of mine I am trying to compare 4 methods (NB,RandomForest,SVM, MLP).  In the process it reaches a step where the data set is multiplied into 4 distinct wrappers (standard Optimize selection) in which the feature selection will take place. In the inner learner of the wrapper I am using the x-validation to obtain the performance of each learner and then obtain the feature set. However I want to be sure that datasets created by the cross-validation (in this case 10) are exactly the same for consistency sake between all the methods, how would I do that?

So summing up, how do I guarantee for each wrapper I have the same rows within folds each time.
« Last Edit: June 27, 2014, 05:06:31 PM by Salo » Logged
Jr. Member
Posts: 93

« Reply #1 on: June 27, 2014, 08:48:15 PM »

Activate "Use local random seed" in the X-Validation operator
Pages: [1]
Jump to: