I have a table of 140 features and try to predict binary classification.

I create two dimensions out of this for visualization with PCA. In this visualization I can see, that data is not separable but overlay almost perfectly.

enter image description here

Can I create from this any conclusions? Is this mean that the target value can't be predicted?

  • $\begingroup$ Be careful with the language in your post. Did you mean to say the data isn't linearly seaprable? The donut in the bottom left looks separable but not linearly separable. Also, what if you trained a knn classifier with this data? If the data is nicely chucked as the image suggests it may do well. Thirdly, be real careful when looking at visualizations like this. Just to make sure any weirdness isn't happening with plotting I would create two plots (one for each class). $\endgroup$
    – juicedatom
    Apr 20 at 15:47

Your Answer

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

Browse other questions tagged or ask your own question.