Pearson's correlation function is too slow.

[erkghlerngm44]

import timeit

a = [1,1,2,4,4,2,6,3,5,3,6,8,5,9,3,1,5,1,8,3,7,4,7,3,7,2,8,8,9,9,9,6,5,3,2]
b = [6,4,7,9,3,1,5,2,7,3,8,4,9,1,2,4,8,5,2,1,2,8,2,9,1,2,4,6,3,5,6,8,4,2,1]

timeit.timeit("malaffinity.calcs.pearson(a, b)", setup="import malaffinity; from __main__ import a, b", number=100)
# 1.3594748531468213

timeit.timeit("scipy.asscalar(scipy.stats.pearsonr(a, b)[0])", setup="import scipy, scipy.stats; from __main__ import a, b", number=100)     
# 0.006378829013556242

Pretty big difference, and it's really noticeable when you're calculating affinity with thousands of people.

Options:

• Re-add scipy as a dep and use that
  • Too slow to import, a nightmare to install, etc. Bad option.
• Make malaffinity.calcs.pearson faster by rewriting it.
  • Nick the code from scipy and use that?
  • Will need to find a fast and neat implementation of Pearson's.