planning 2018.05.16

info

• notebooks to be merged without any serious changes
• name is gradus

status

• implement make_transformer()
• Unintuitive adapter syntax
  • the same notation for input data and adapters?

MR results for:

• API for ensembling (MR)

discussion

• define single caching directory for all transformers (Per project? Context handler? Something else?) -> do not do it per transformer
• cache/save output
• delete cache at the end of pipeline computations
• separate output dirs for train validation test and user-defined splits
• Input has complicated notation: nested dicts in input -> simplify interface -> DataStep should merge input_step and input_data into one API piece.

data = {'input':
          {
               'X': X_train,
               'y': y_train,
           }
        }

notes

• You add input that is never used -> you see it on the graph.
• cache_transformer -> persisted_transformer
• Do not transform twice!