I have been using matplotlib as a replacement for MatLab and
> have had a great deal of success repicating basic plots. I
> was now trying to replicate some of the more value added
> functionality like the normplot and cdfplot. The normplot is
> particularly tricky because you need to scale the axises by
> the inverse normal cdf. There seems to be functionality to
> write scalers but the only ones that come with matplotlib out
> of the box are linear and log.
> Does anyone know how I might be able to achieve this
Unfortunately, mpl's transformation stuff is in C++, and is not a
terribly extensible design. It would be easy to add new functions to
the SeparableTransformation, but what you need is really a functional,
eg a class that is initialized with a set of values and transforms the
The place to start poking around is src/_transforms.h and
src/_transforms.cpp in the SeparableTransformation class.
I'm out of town for the next week or so and will ruminate on this
while I'm away.