Converter for pyod.models.iforest.IForest

Is it possible to get converter for pyod anomaly detection models? https://pyod.readthedocs.io/en/latest/_modules/pyod/models/iforest.html

pyod Iforest is a Wrapper of scikit-learn Isolation Forest with more functionalities.

Traceback (most recent call last):

  File "<ipython-input-91-f753e8bf7b2f>", line 1, in <module>
    IFOnnx_model = convert_sklearn(model1, initial_types=initial_type)

  File "C:\Users\pl\Anaconda3\lib\site-packages\onnxmltools\convert\main.py", line 66, in convert_sklearn
    custom_conversion_functions, custom_shape_calculators)

  File "C:\Users\pl\Anaconda3\lib\site-packages\skl2onnx\convert.py", line 150, in convert_sklearn
    topology.compile()

  File "C:\Users\pl\Anaconda3\lib\site-packages\skl2onnx\common\_topology.py", line 903, in compile
    self._infer_all_types()

  File "C:\Users\pl\Anaconda3\lib\site-packages\skl2onnx\common\_topology.py", line 756, in _infer_all_types
    operator.infer_types()

  File "C:\Users\pl\Anaconda3\lib\site-packages\skl2onnx\common\_topology.py", line 215, in infer_types
    type(self.raw_operator)))

MissingShapeCalculator: Unable to find a shape calculator for type '<class 'pyod.models.iforest.IForest'>'.
It usually means the pipeline being converted contains a
transformer or a predictor with no corresponding converter
implemented in sklearn-onnx. If the converted is implemented
in another library, you need to register
the converted so that it can be used by sklearn-onnx (function
update_registered_converter). If the model is not yet covered
by sklearn-onnx, you may raise an issue to
https://github.com/onnx/sklearn-onnx/issues
to get the converter implemented or even contribute to the
project. If the model is a custom model, a new converter must
be implemented. Examples can be found in the gallery.

sample code for IForest

import pandas as pd
from pyod.models.iforest import IForest
from sklearn.preprocessing import MinMaxScaler

data1 = {'First':  [500,500,400,100,200,300,100],
         'Second': ['a','b','a','b','a','b','c'],
         }
df1 = pd.DataFrame(data1,columns=['First','Second'])
dumdf1 = pd.get_dummies(df1) 
scaler = MinMaxScaler()
scaler.partial_fit(dumdf1)
sc_data = scaler.transform(dumdf1)
model1 = IForest(n_estimators=10,
               bootstrap=True,
               behaviour='new',
               contamination=0.1,
               random_state=np.random.RandomState(42),
               verbose=1,
               n_jobs=-1).fit(sc_data)
feature_names2 = dumdf1.columns

from onnxmltools.convert.common.data_types import FloatTensorType
from onnxmltools import convert_sklearn

initial_type = [('float_input', FloatTensorType([None, len(feature_names2)]))]
IFOnnx_model = convert_sklearn(model1, initial_types=initial_type)

Thanks in advance

[xadupre]

Thanks for letting me know. I’ll close the issue. Feel free to reopen it if something fails again.

[xadupre]

There are many extensions to scikit-learn and sklearn-onnx cannot support all of them. You may follow the tutorial to implement a new converter Implement a new converter. I could also add an example to the documentation to show how to do it. What do you prefer?