snae/nutrient_models

There are several .pkl files in this directory. Below their meaning is explained.

• A model can be rf (random forest) or xgb (XGBoost).
• For NEVO, there is a model trained on the dataset's own columns and on columns that are mapped to FNDDS.
• Most models are trained on six nutrients, meaning the six basic nutrients. A reference of all means trained on all available nutrient columns. This differs per dataset and therefore per model.
  • Considering using the model on a different dataset, six is probably easiest, as column names are sugars_100g', 'proteins_100g', 'fat_100g', 'carbohydrates_100g', 'sodium_100g' . If not using six`, you can infer the necessary columns from the model, as it will ask you for the exact same names.
• The name of the dataset is used to indicate on which dataset the model is trained, such as leda or off.

oneclf can be ignored.