Cosine correlation
Cosine correlation is a common similarity measure when doing hierarchical clustering.
The Cosine correlation between two points, a and b, with k dimensions is calculated as:
where
The cosine correlation ranges from +1 to -1 where +1 is the highest correlation. Complete opposite points have correlation -1.
Comparison between Cosine correlation and Correlation
The difference between Cosine correlation and Correlation is that the
average value is subtracted in Correlation. In the example below, the Cosine
correlation will be +1 between any combination of points
a,
b, and
c, but it will be slightly less than that between point d and
any of the other points (+0.974). However, the regular Correlation will be +1
between any of the points, including point
d.
