I've been reading about Fisher's Linear Discriminant Analysis lately, and I noticed that the objective function (particularly for two-class classification) to be maximized contains scatter terms instead of variance, in the denominator. Why is that?

To clarify, the scatter of a sample is just the variance multiplied by the number of data points in the sample.

Thank you!

  • $\begingroup$ I think this may be a better fit at stats.stackexchange.com - I'm not 100% sure, but I think the answer will be to do with how variance of dataset mean varies depending on variance of individual terms. There are far more experts in this kind of pure statistics over at Cross Validated, and LDA is only weakly related to AI and ML. $\endgroup$ – Neil Slater Jul 14 at 8:15
  • $\begingroup$ I think you may find the answer here: stats.stackexchange.com/questions/123490/… $\endgroup$ – Daniel Wiczew Jul 19 at 18:05

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.