ONNX New Converters#

mlprodict implements a couple of converters. The function register_converters registers them and makes them visible to sklearn-onnx so that a pipeline including one of the supported operators can be fully converted.

<<<

from mlprodict.onnx_conv.register import register_converters
from pandas import DataFrame
from pyquickhelper.pandashelper import df2rst
models = register_converters()
data = [(cl.__name__,
         dict(name=cl.__name__,
              module=":epkg:`{}`".format(cl.__module__.split('.')[0])))
        for cl in models]
data.sort()
data = [_[1] for _ in data]
df = DataFrame(data)
print(df2rst(df))

>>>

name	module
Booster	lightgbm
CustomScorerTransform	mlprodict
LGBMClassifier	lightgbm
LGBMRegressor	lightgbm
TransferTransformer	mlinsights
WOETransformer	skl2onnx
WrappedLightGbmBooster	mlprodict
WrappedLightGbmBoosterClassifier	mlprodict
XGBClassifier	xgboost
XGBRegressor	xgboost

scikit-learn Converters and Benchmarks

Visual Representation of scikit-learn models