What is the algorithm used by Matlab to create the Weighted KNN method on the classification App?

I am using the classification app on a dataset and getting a good acuracy. But I would like to use a distance measured not offered by Matlab under that option (dtw). Since I cannot change the Matlab file, I would like to know at least where they got their algorithm from or how it works so I can write a script myself and use the distance measure I want. Can anybody help?

Answer 1

[Mark Levin]

I have the same question... It looks to me like 1. Normalize your point 2. find the distance to 10 NN 3. take the 1./dist^2 sum it for every class and take the biggest... and don't forget to use apriori probability...