"Kulsinski" dissimilarity seems wrong (Trac #1484)

Original ticket http://projects.scipy.org/scipy/ticket/1484 on 2011-07-28 by trac user muellner, assigned to trac user peridot.

The "Kulsinski" dissimilarity in the module scipy.spatial.distance seems wrong.

First, the function does not do what the documentation

says: it always returns integer values. Maybe this is a forgotten float(...) in scipy/spatial/distance.py, line 421.

Second, the definition does not agree what I find elsewhere on the net. Compare

and

Can someone please check whether the SciPy definition makes sense and/or is intenionally different from the definition found in R?

Provide feedback