My recent work on splines did indeed give improved results on simple test data sets, but I wondered what would happen for unusual cases. Here is a trivially small example data set where scipy's Univariate Spline does not work:
6.1 1
6.1 2
8.1 3
8.1 4
Note that there are four data points, but only two distinct values on the X axis, 6.1 and 8.1. Try this on the site and the spline silently fails. It also silently fails in my simple Python script spline tests.
My thought is to remove scipy splines from the site entirely as they are not working and fail silently.
James