scipy interp function has been removed as of v1.12

[mooreryan]

(I think that's the version it was removed at least.)

In Submodule 4, Tutorial 1, the ROC curve section uses interp from scikit learn. This has been removed in recent scipy versions, so we need to replace it with something else.

There is numpy.interp, which may be a direct replacement. (What do you think @chenchuming ?)

There is also interp1d from scipy.interpolate. It definitely has a different interface, but maybe it is doing what you want Chuming.

Potentially something like this:

from scipy.interpolate import interp1d
# ...
interpolation = interp1d(fpr[i], tpr[i])
mean_tpr += interpolation(all_fpr)

[chenchuming]

It looks good to me.

[mooreryan]

Since we're likely going with the conda set up, I can do like I did in #46 and figure out which version of sklearn to use so that we don't have to change this.