pandas as_matrix deprecation

JGCRI / pygcam

Python tools for automating GCAM workflows and managing experiment configuration

Other

24 stars 11 forks source link

pandas as_matrix deprecation #33

Open imanyakin opened 2 years ago

imanyakin commented 2 years ago

Trying to run MCS analysis, but gt complains:

'DataFrame' object has no attribute 'as_matrix'

This appears to have been deprecated since 0.23.0.

There are only two instances of as_matrix use in pygcam/mcs/sensitivity.py
seems like they can be fixed with relative ease:

'bounds' : data.as_matrix(columns=['low','high']), -> 'bounds' : data[['low','high']].to_numpy(),

self.results = self.resultsDF[resultName].as_matrix() -> self.results = self.resultsDF[resultName].to_numpy()

rjplevin commented 2 years ago

As you have discovered, the entire set of code related to SALib is very outdated. I haven't used it or worked on it in years. I cannot recommend using anything other than the default Monte Carlo methods at this time.

Please understand that I have no funding for further work on pygcam, so the only features that get added at this point are those that I require for my own work. I use pygcam in all my ongoing work with GCAM, but I do not use every feature, and I work only on the Mac and Linux platforms. I rarely do testing on Windows. My current work involves a custom version based on GCAMv5.3, and I haven't had opportunity to use 6.0 yet.

I'd be happy to develop / complete new features if funding were available.

imanyakin commented 2 years ago

Hi Richard, yes of course - sorry I didn't mean to 'create work'