Currently, the final anchor element prediction score is computed by multiplying different anchor metric values with integer scales. This means that changing the scales results in changing the anchor element score. This makes it difficult to experiment with different scales for different anchor element metric values.
We should normalize the final prediction score over a scale of 100. This would allow us to experiment with using different scales, and compare the accuracy of different prediction algorithms.
Comment 1 by bugdroid1@chromium.org
, Oct 4