Giters

interpretml / interpret-community

Interpret Community extends Interpret repository with additional interpretability techniques and utility functions to handle real-world datasets and workflows.

Home Page:https://interpret-community.readthedocs.io/en/latest/index.html

Repository from Github https://github.cominterpretml/interpret-communityRepository from Github https://github.cominterpretml/interpret-community

name 'LGBMRegressor' is not defined

anindya5 opened this issue · comments

commented

I am getting this error while trying to use the MimicExplainer in your advanced-feature-transformations-explain-local.ipynb notebook.

I have lightgbm and all installed, but cannot seem to get this working. The SHAP based explainers are working fine, but running

explainer = MimicExplainer(clf.steps[-1][1], 
                            x_train, 
                            LGBMExplainableModel, 
                           augment_data=True, 
                            max_num_of_augmentations=10, 
                            features=x_train.columns, 
                            transformations=transformations, 
                            allow_all_transformations=True)

gives the error

NameError                                 Traceback (most recent call last)
<ipython-input-27-abf3dd406d1c> in async-def-wrapper()
     27 
     28 
---> 29 
     30 
     31 

~/anaconda3/envs/interp/lib/python3.6/site-packages/interpret_community/dataset/decorator.py in init_wrapper(self, model, initialization_examples, *args, **kwargs)
     36         if not isinstance(initialization_examples, DatasetWrapper):
     37             initialization_examples = DatasetWrapper(initialization_examples)
---> 38         return init_func(self, model, initialization_examples, *args, **kwargs)
     39     return init_wrapper

~/anaconda3/envs/interp/lib/python3.6/site-packages/interpret_community/mimic/mimic_explainer.py in __init__(self, model, initialization_examples, explainable_model, explainable_model_args, is_function, augment_data, max_num_of_augmentations, explain_subset, features, classes, transformations, allow_all_transformations, shap_values_output, categorical_features, model_task, reset_index, **kwargs)
    302             explainable_model_args[ExplainParams.SHAP_VALUES_OUTPUT] = shap_values_output
    303         self.surrogate_model = _model_distill(self.function, explainable_model, training_data,
--> 304                                               original_training_data, explainable_model_args)
    305         self._method = self.surrogate_model._method
    306         self._original_eval_examples = None

~/anaconda3/envs/interp/lib/python3.6/site-packages/interpret_community/mimic/model_distill.py in _model_distill(teacher_model_predict_fn, uninitialized_surrogate_model, data, original_training_data, explainable_model_args)
     67                                                         **explainable_model_args)
     68     else:
---> 69         surrogate_model = uninitialized_surrogate_model(**explainable_model_args)
     70     if is_classifier and teacher_y.shape[1] == 2:
     71         # Make sure output has only 1 dimension

~/anaconda3/envs/interp/lib/python3.6/site-packages/interpret_community/mimic/models/lightgbm_model.py in __init__(self, multiclass, random_state, shap_values_output, classification, **kwargs)
     93             initializer = LGBMClassifier
     94         else:
---> 95             initializer = LGBMRegressor
     96         self._lgbm = initializer(random_state=random_state, **initializer_args)
     97         super(LGBMExplainableModel, self).__init__(**kwargs)

NameError: name 'LGBMRegressor' is not defined
commented

Hi @anindya5 --

This appears to be a question that's better suited for the interpret-community repo. Transferring it there so they can help you.

-InterpretML

commented
commented

@anindya5 strange, can you do the import manually in your python environment?

from lightgbm import LGBMRegressor, LGBMClassifier, Booster
init_func = LGBMRegressor

Perhaps the import errors are getting suppressed here, do you see anything printed:

with warnings.catch_warnings():
    warnings.filterwarnings('ignore', 'Starting from version 2.2.1', UserWarning)
    import shap
    try:
        from lightgbm import LGBMRegressor, LGBMClassifier, Booster
        import lightgbm
        if (version.parse(lightgbm.__version__) <= version.parse('2.2.1')):
            print("Using older than supported version of lightgbm, please upgrade to version greater than 2.2.1")
    except ImportError:
        print("Could not import lightgbm, required if using LGBMExplainableModel")
commented

@anindya5 is this issue resolved? Are you able to proceed?

Regards

commented

@anindya5, closing this issue since there is no response on this thread. Please feel free to reopen if you continue to run into this error.

Regards

commented

Still seeing this issue ? Does anyone has any resolution ?

commented

@nayanaramakanth Sorry about the issue you are having. Do you have lightgbm package installed in your environment? Can you show the output of:

pip show lightgbm

Specifically what version of lightgbm you have?

commented

Hi,

I'm having the same problem. In my case I'm following this AzureML tutorial.
https://docs.microsoft.com/en-us/azure/machine-learning/how-to-machine-learning-interpretability

And when I try to execute this part:

from interpret.ext.blackbox import MimicExplainer

from lightgbm import LGBMRegressor, LGBMClassifier, Booster
init_func = LGBMRegressor

# you can use one of the following four interpretable models as a global surrogate to the black box model

from interpret.ext.glassbox import LGBMExplainableModel
from interpret.ext.glassbox import LinearExplainableModel
from interpret.ext.glassbox import SGDExplainableModel
from interpret.ext.glassbox import DecisionTreeExplainableModel

# "features" and "classes" fields are optional
# augment_data is optional and if true, oversamples the initialization examples to improve surrogate model accuracy to fit original model.  Useful for high-dimensional data where the number of rows is less than the number of columns.
# max_num_of_augmentations is optional and defines max number of times we can increase the input data size.
# LGBMExplainableModel can be replaced with LinearExplainableModel, SGDExplainableModel, or DecisionTreeExplainableModel
explainer2 = MimicExplainer(model, 
                           x_train, 
                           LGBMExplainableModel, 
                           augment_data=True, 
                           max_num_of_augmentations=10, 
                           features=breast_cancer_data.feature_names, 
                           classes=classes)_

I have the commented error

C:\ProgramData\Anaconda3\lib\site-packages\interpret_community\mimic\models\lightgbm_model.py in __init__(self, multiclass, random_state, shap_values_output, classification, **kwargs)
    173             initializer = LGBMClassifier
    174         else:
--> 175             initializer = LGBMRegressor
    176         self._lgbm = initializer(random_state=random_state, **initializer_args)
    177         super(LGBMExplainableModel, self).__init__(**kwargs)

NameError: name 'LGBMRegressor' is not defined_

Here is the result of "pip show lightgbm"

Name: lightgbm
Version: 3.3.2
Summary: LightGBM Python Package
Home-page: https://github.com/microsoft/LightGBM
Author: None
Author-email: None
License: The MIT License (Microsoft)
Location: c:\programdata\anaconda3\lib\site-packages
Requires: numpy, scipy, wheel, scikit-learn
Required-by:

Thanks in advance for your support

commented

Hi @nayanaramakanth can you help with this, thanks in advance....

commented

@rnavarromatesanz
can you try doing an import in your python environment manually, eg:

from lightgbm import LGBMRegressor, LGBMClassifier, Booster

this error indicates that this import is just not working for some reason