Dataset building/encoding

[FAMILIAR-project]

Hi,

The situation is getting better with dataset_after_encoding.csv but we need to improve/generalize the solution.

Here https://github.com/TuxML/tuxml-datasets are different CSVs of the database.
In fact config_bdd30-100.pkl has been obtained through the merging of CSV (see TUXML-csv-building.ipynb)

• remove "sizes" columns (to save some spaces and ease the processing: we don't need it for this analysis)
• write a script to encode True/False values into 0, 1, 2 (as did for the dataset 30-100) and export as CSV
• merge the whole data into an unique CSV