Kevin Sheppard

Code is pretty easy, and now this is documented.

I think that we should probably add a function that would handle the obvious cases.

These aren't really maintained. It looks like the underlying notebooks were removed. Not ised where these notebooks were stored though since it isn't easy to resolve.

Or perhaps something happened with nb viewer, which seems more likely.

Normally transforming forecasts is not trivial. For example if you take the Ln of your data, then when you forecast you need to that exp(fcast+sigma^2/2) where sigma is the std...

In general you need to handle the retransform yourself.

Lots of SciPy instability these days. Failure in descriptive was also SciPy related fro changes in mode.

The shapiro test should just be cut. It is only testing SciPy code.

SciPy seems to have been fixed.

The threads are in your BLAS library. If you are using MKL, then you, before your session ``` export MKL_NUM_THREADS=1 # Or whatever ```` If using pip-NumPy ``` export OPENBLAS_NUM_THREADS=1...