1
$\begingroup$

I noticed that in my dataset a particular column is highly separable where it splits the data perfectly into 5 distinct classes (re-evaluated where class2 means better than class1). I would like to study the underlying structure using a clustering model from this same data set.

  1. Should I include this column as a variable for clustering despite knowing that this feature is highly separable?

  2. Would this feature create any bias or affect the results for the clustering model?

All these with the assumption that I will be using a K-means Algorithm

$\endgroup$
3
  • $\begingroup$ Feature engineering means manipulating raw features to derive better features (maybe in lower dimensionality) which nevertheless describe better the data. It is the case that all necessary information for a certain task may be only in a sub-manifold or the original manifold of data. $\endgroup$
    – Nikos M.
    Nov 25, 2021 at 17:15
  • $\begingroup$ Please let me know if you are satisfied with the answer? If not I will try my best possible way to edit it. $\endgroup$ Nov 28, 2021 at 3:55
  • $\begingroup$ NikosM and DevashishPrasad with regards to your answers, i think it is great! I see, so to put both explanations together, the idea of feature engineering can help describe the data points in lower dimensionality, such that it would / might be more separable which will help the clustering algorithm and in which case, It is then preferred to use a evaluation method to study the underlying patterns. $\endgroup$
    – Cosq
    Dec 6, 2021 at 12:30

1 Answer 1

0
$\begingroup$

Would this feature create any bias or affect the results for the clustering model?

This is all you need to think about while performing clustering. You can evaluate your clustering algorithm to assess whether it is performing well or not. And this question has lots of resources that will help you how one can evaluate clustering algorithms. In this way, you will be able to evaluate your clustering algorithm and also analyze the effects of various features on your algorithm.

$\endgroup$

Your Answer

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

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