liuchbryan / generalised_forest_tuning Goto Github PK

Update the code which generate the figures to deliver greater clarity.

• Code for downloading and installing anaconda
• Code for creating a new environment in anaconda (so that it would not mess up with users' existing python installation)
• Code for installing required libraries (pandas, numpy, scikit-learn, matplotlib, gcc, pybo, jupyter), include force update for libgfortran (see ContinuumIO/anaconda-issues#686 (comment))
• Code to load public dataset into local_resources directory
• Some pointer in top-level README.md file

As titled, so that code for prerequisite setup (e.g. library setup, loading data to prespecified folders) knows where they are loading stuffs into.

Migrate the:

• Python files
• Jupyter notebooks
• Data files (the small ones, e.g. grid search AUC/RMSPD/cost)

which contains the experiments referenced in the paper.

Python 2 will be sunset on 1 Jan 2020. This issue is to gauge the demand in upgrading the research code to Python 3.

The mathematics has not changed, and hence implementing the formulas from scratch shouldn't be the toughest job, any collaboration is welcomed (and much appreciated) though!

Provide the complete derivation document that RMSPD can be expressed as the empirical estimate of interaction between variance of model and covariance of sampling.