Config.yaml should allow mapping of Feature column to Datatype mapping

[rayleeriver]

From Auger.ai console, I can specify a specific feature column to specific DataType, e.g. xyz->Categorical. Same functionality is missing for A2ml.

[skatedplrn]

Please note that most of the type discovery happens automatically, if you need fine tuning: we support special settings for experiment: categoricals - list of categoricals columns label_encoded - list of columns which should be label encoded, if categorical values > 50 then columns assigned to label encoding automatically date_time - list of datetime columns time_series - time series columns, used only if model_type is timeseries

Should update doc anyway

[rayleeriver]

How about changing a column mapping from Double to Integer etc?

[adamblum]

hey @rayleeriver - we arent going to tackle such things that are effectively dataprep (to run with azure we would have to change the data for this). but if you want to tackle this would be appreciated. in the meantime we will leave open for future release