Currently, the implementation of Cohen's weighted kappa returns unnormalized scores
for observed and expected disagreement. This still yields the correct kappa score,
but prevents easy interpretation and analysis of intermediate values.
TODO:
* Add test case demonstrating the issue
* Normalize the scores by d_max
Originally reported on Google Code with ID 7
Reported by
chmeyer.de
on 2014-08-20 16:11:54