I have set of 13 dimensional vectors split into three classes, and I'm building a classifier using supervised learning.

After running c = Classify[..] on Mathematica 11.0 and evaluating ClassifierInformation[c], Mathematica replies

Method                      K-nearest neighbours
Number of classes           3
Number of features          1
Number of training examples 6479
Number of neighbors         5
Distance function           EuclideanDistance

The input to the classifier were vectors all 13 long. Am I right in taking the reply above as saying that only one of the vector dimensions was used to achieve the classification?

  • $\begingroup$ Can you present your code with a lesser number of training examples (say 20-30 items)? $\endgroup$ – user64494 Apr 10 at 14:33

Let us consider an example from ?Classify:

d = Dataset[{<|"age" -> 32, "height" -> 160, "gender" -> "female"|>,
<|"height" -> 183, "age" -> 41, "gender" -> "female"|>,
<|"height" -> 123, "age" -> 30, "gender" -> "female"|>,
<|"height" -> 175, "age" -> 21, "gender" -> "male"|>,
<|"height" -> 150, "age" -> 11, "gender" -> "male"|>,
<|"age" -> 52, "height" -> 164, "gender" -> "female"|>}];
c = Classify[d -> "gender"];

{"age", "height"}

ClassifierInformation[c,  "FeatureNumber"]


Hope this sheds light on the topic under consideration.

  • $\begingroup$ I’m afraid your point is not at all clear to me. Can you explain why this should be an explanation of the behavior OP sees? $\endgroup$ – MarcoB Apr 11 at 13:01
  • $\begingroup$ @MarcoB: Because OP does not present any Mathematica code, I decided to demonstrate "FeatureNames" and "FeatureNUmber" in the above example, no more and no less. Unfortunately,OP still does not give any response to my request and anwser though Op was present at the forum two hours ago. I believe civilized people use to give feedbacks. $\endgroup$ – user64494 Apr 11 at 15:12

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.