Guys,
I'm performing kNN prediction of numerical variable and I have some questions:
1. How exactly the weightning in WEKA IBk is performed? I am using euclidean distance and 1/distance as weights but how this is applied?
2. How to get information for method evaluation which weka predictor calculate (e.g. r, mae...)?
3. What type is the calculated correlation coefficient - Pearson, determination, squared version of any of these?