To avoid the warning, I changed a line in /path/to/scipy/integrate/quadrature.py L356.
shapex = ones(nd)
becomes
shapex = ones(nd, dtype='int64')
The problem is that np.ones by default creates a float64 array, and now numpy warns when floats are used as indices, so casting as ints upon creation avoids the problem. This probably appears in a bunch of places in scipy and I haven't tried to fix it anywhere else.
I traced this issue to scipy.integrate.simps which is called in tripDistances.avgMillisFromBlock
It appears to be a warning new in numpy 1.8, so may be version specific (e.g. http://stackoverflow.com/questions/20084218/deprecation-warning-in-scikit-learn-svmlight-format-loader)
To avoid the warning, I changed a line in /path/to/scipy/integrate/quadrature.py L356. shapex = ones(nd) becomes shapex = ones(nd, dtype='int64')
The problem is that np.ones by default creates a float64 array, and now numpy warns when floats are used as indices, so casting as ints upon creation avoids the problem. This probably appears in a bunch of places in scipy and I haven't tried to fix it anywhere else.