1
$\begingroup$

i have textual data points.. (1000+) text documents.. i used K-means to cluster these data, then used SVD method upon (TF-IDF matrix) then select the 2nd and 3d columns in VT matrix to represent these documents in 2D in scatter plot, so after i plotting points and colored them based on their cluster id , i found there are many points are overlapping with others... so kindly i would like to know is there any problem in clustering algorithm or scatter plot do not reflect the correct nature of data, or should i use another method to plotting my data points.

thanks ... attached my result data points in scatter plot where each color is an cluster id. enter image description here

$\endgroup$
  • $\begingroup$ Why not the first column? $\endgroup$ – Anony-Mousse Jul 21 '17 at 22:22
1
$\begingroup$

The scatter plot is only showing two dimensions of a high-dimensional space, so it's expected that there will be overlap. That is, some points that are different in n-dimensional space will look similar in a 2-dimensional space.

$\endgroup$

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.