Closed daveesterman closed 11 years ago
Euclidian distance score should not be rescaled to [-1,1]. This does not have the same meaning as the range [-1, 1] when returned as Pearson correlation coefficient. With Pearson, negative correlation means "as x increases, y certainly decreases". with Euclidian, negative score mean "points are relatively far apart".
The range returned by Euclidian distance is now (0,1].
Potential to divide by 0 in Filter.getNormalizedWeightedScore