Solaris

Requirements

pandas>=0.12.0 numpy>=1.7.1 docopt>=0.6.1 netCDF4>=1.0.4

Data

All data files are expected to be found in the data subfolder. Each dataset is a joblib pickled dict holding training and test set as solaris.sa.StructuredArray.

To create this format from the Kaggle download use the script solaris.load_data:

$ python -m solaris.load_data

This will save the transformed data into the data subfolder. The name is data/data.pkl.

To generate the interpolated blocks use the script solaris.kringing.

$ python -m solaris.kringing