I am developing a k-means clustering algorithm, and I have obtained the ideal number of clusters based on the elbow method. However, despite the fact that the error diminishes a lot with the number of clusters, when it stabilizes the SSE value is still in the order of 2x10^+6. Is this acceptable? I can't seem to find any information on this.

Thank you in advance!


There is no benchmark for an acceptable SSE.

Assume your data are points located in two-dimensional space. If you measure distances in millimeters, in meters or in kilometers will change the SSE by factors of $10^6$, regardless of the clustering.

What is "acceptable" will depend on your problem, your data and alternatives to the clustering you have chosen.


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.