mz2 / swot Goto Github PK

It is not currently possible to map categorical fields into categorical fields with the named column dimension mapper (no -addCategoryWithName: values: or -assignCategoryWithName:toColumnWithIndex: calls are made by it to the mapped dataset it outputs).

Currently only label field mapping is supported (label field translated as is, with a potential column offset depending on its column index).

This is not an immediate need because categorical fields would need to be translated to floating point valued fields (using ALGLIB like translation into multiple effectively binary valued fields, for instance categorical value with four values A,B,C,D could get translated into four columns, with values [1,0,0,0], [0,1,0,0], [0,0,1,0], [0,0,0,1]) for all of the learning algorithms currently envisaged for inclusion in Swot.