6
$\begingroup$

Is there a way to create my own ClassifierFunction? I would like to combine predictions from multiple independent classifiers into a single ensemble ClassifierFunction and be able to use things like ClassifierMeasurments etc. Afaik, it seems like the only way to get a ClassifierFunction is via Classify or NetTrain family of functions.

$\endgroup$
6
  • $\begingroup$ +1 from me. It's possible to combine classifiers/predictors. But this is undocumented functionality. I have decided to open the new question because your should be rewritten completely. $\endgroup$ Dec 11, 2016 at 15:12
  • $\begingroup$ @AlexeyGolyshev, why should my question be rewritten completely? $\endgroup$
    – Abel Brown
    Dec 11, 2016 at 18:02
  • $\begingroup$ Your question is unclear on the first sight and was downvoted by someone. I thought to modify your text, add code and images. But what if I am wrong and your original question is a little different. For example, I ensemble my models this way: c1=Classify[...];c2=Classify[...];ensemble[x_]:=(c1[x]+c2[x])/2. Simple average or weighted average. You can create ConfusionMatrix. See this question and answer + Riffle[ensemble[x],y]//Partition[#,2]&//Tally $\endgroup$ Dec 12, 2016 at 8:07
  • $\begingroup$ @AlexeyGolyshev I'm not sure my question could be more clear. Thanks for your code snippet but, I'm not interested in generating confusion matrices. Also, ensemble[x_]:=(c1[x]+c2[x]‌​)/2 does not generate a ClassifierFunction and therefore can not be used as input to ClassifierMeasurements. $\endgroup$
    – Abel Brown
    Dec 13, 2016 at 13:16
  • $\begingroup$ Abel, I have found solution! $\endgroup$ Jan 12, 2017 at 15:22

0

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.