Investigate options for InterRowMSAS #669
Description
Problem Description
Right now, the InterRowMSAS metric takes the direct difference between a value in row n and row n+1. Then, it averages out all these differences. As a result, the computation effectively cancels out all terms besides the first and last
(row 2 - row 1) + (row 3 - row 2) + (row 4 - row 3) + ... + (row n - row n-1)
= row n - row 1
I'm filing this issue to track whether there is a different form of computation that would be more appropriate for this metric. Alternatives:
- Do not average out the differences between each sequence. Instead, add the differences to an overall distribution D_r or D_s.
- (Similar to taking a log) Apply a transform each number. Eg. Squaring all values, and identifying the square root of the differences, eg.
sqrt((r+x)**2 - (r)**2)