Closed laurent-thiebaud-gisaia closed 4 years ago
We should dispatch transformers into the following sub-packages :
ml
: transformers that requires ARLAS-ml stufftimeseries
: transformers that deals with states and idsfeatures
: transformers that compute new features (distance, speed, ...)tools
: transformers that perform basic operations on the dataframe (value replacement, colonne mapping, ...)misc
: other transformers
The transformers folder is get messy. Transformers within the same hierarchy should be put in dedicated folders. Transformers that are customer-specific should be put in the customer specific project (this may be checked with data scientists).