The TD_ColumnTransformer function transforms the input table columns in a single operation. You only need to provide the FIT tables to the function, and the function runs all transformations that you require in a single operation.
The function performs the following transformations:
- TD_Scale Transform
- TD_Bincode Transform
- TD_Function Transform
- TD_NonLinearCombine Transform
- TD_OutlierFilter Transform
- TD_PolynomialFeatures Transform
- TD_RowNormalize Transform
- TD_OrdinalEncoding Transform
- TD_OneHotEncoding Transform
- TD_SimpleImpute Transform
TD_ColumnTransformer supports multiple instances of same FitTable for NonLinearCombineFitTable. For all other types, only single instance is allowed.
You must create the FIT tables before using the function and you must provide the FIT tables in the same order as in the training data sequence to transform the dataset. The FIT tables must have a maximum of 128 columns.
The TD_BincodeFit function has a maximum of 5 columns when using the variable-width method.