0
$\begingroup$

I wanted to cluster data points into 2 clusters, I am using clustmix package from R I wanted to understand importance of each of the feature, I have 203 features. I have tried featureimp package from R , but it's more focussed around using misclassification rate to define importance. Is it a correct way to identify the feature importance of not

Any suggestions are invited

Improve this question
$\endgroup$
2
  • $\begingroup$ Hi @ashishpople, welcome to the site. You could train a logistic regression classifier using the clusters as labels, and then compute the predictor importance, e.g. with the varImp function. $\endgroup$
    – noe
    Commented Jan 9 at 8:48
  • $\begingroup$ For a complete tutorial on logistic regression, including variable importance, you can check here. $\endgroup$
    – noe
    Commented Jan 9 at 9:58

0

Reset to default

Your Answer

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