1
$\begingroup$

I'm working on a classification (3 classes) of unbalanced weather data having 22 features. Even after applying PCA and t-SNE the data is overlapping. The best classification score achieved so far is using the tree-based method. What can be the reason for such PCA plot and what techniques I can implement to get more than 80% accuracy? pca ploty-sne plot

Improve this question
$\endgroup$
1
  • 1
    $\begingroup$ It could be that your features are not terribly predictive of the outcome. Worth reading… $\endgroup$
    – Dave
    Commented Oct 26, 2021 at 0:19

1 Answer 1

Reset to default
1
$\begingroup$

I would recommend to use UMAP. It is a superior algorithm to both of them:

Improve this answer
$\endgroup$

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.