search

oegedijk / explainerdashboard

Quickly build Explainable AI dashboards that show the inner workings of so-called "blackbox" machine learning models.
http://explainerdashboard.readthedocs.io
MIT License
2.32k stars 332 forks source link

ImportError: cannot import name 'dtreeviz' from 'dtreeviz.trees' #282

Closed lalitsc12 closed 11 months ago

lalitsc12 commented 1 year ago

I am getting this error not sure what is the issue.

ImportError: cannot import name 'dtreeviz' from 'dtreeviz.trees'

i am using the simple example in jupyter notebook

from pycaret.datasets import get_data data=get_data('juice') from pycaret.classification import * s=setup(data,target='Purchase',session_id=123) lr=create_model('lr')

dashboard(lr,display_format='inline')

this is when i get error:. Cell In[7], line 1 ----> 1 dashboard(lr,display_format='inline')

File ~/anaconda3/envs/Lalit_env/lib/python3.9/site-packages/pycaret/utils/generic.py:965, in check_if_global_is_not_none..decorator..wrapper(*args, *kwargs) 963 if globals_d[name] is None: 964 raise ValueError(message) --> 965 return func(args, **kwargs)

File ~/anaconda3/envs/Lalit_env/lib/python3.9/site-packages/pycaret/classification/functional.py:2997, in dashboard(estimator, display_format, dashboard_kwargs, run_kwargs, kwargs) 2944 @check_if_global_is_not_none(globals(), _CURRENT_EXPERIMENT_DECORATOR_DICT) 2945 def dashboard( 2946 estimator, (...) 2950 kwargs, 2951 ): 2952 """ 2953 This function generates the interactive dashboard for a trained model. The 2954 dashboard is implemented using ExplainerDashboard (explainerdashboard.readthedocs.io) (...) 2995 ExplainerDashboard 2996 """ -> 2997 return _CURRENT_EXPERIMENT.dashboard( 2998 estimator, display_format, dashboard_kwargs, run_kwargs, **kwargs 2999 )

File ~/anaconda3/envs/Lalit_env/lib/python3.9/site-packages/pycaret/classification/oop.py:3419, in ClassificationExperiment.dashboard(self, estimator, display_format, dashboard_kwargs, run_kwargs, **kwargs) 3416 dashboard_kwargs = dashboard_kwargs or {} 3417 run_kwargs = run_kwargs or {} -> 3419 from explainerdashboard import ClassifierExplainer, ExplainerDashboard 3421 le = get_label_encoder(self.pipeline) 3422 if le:

File ~/anaconda3/envs/Lalit_env/lib/python3.9/site-packages/explainerdashboard/init.py:1 ----> 1 from .explainers import ClassifierExplainer, RegressionExplainer 2 from .dashboards import ExplainerDashboard, ExplainerHub, InlineExplainer

File ~/anaconda3/envs/Lalit_env/lib/python3.9/site-packages/explainerdashboard/explainers.py:25 21 from pandas.api.types import is_numeric_dtype, is_string_dtype 23 import shap ---> 25 from dtreeviz.trees import ShadowDecTree, dtreeviz 27 from sklearn.pipeline import Pipeline 28 from sklearn.metrics import roc_auc_score, accuracy_score, f1_score

ImportError: cannot import name 'dtreeviz' from 'dtreeviz.trees

snaunheim commented 1 year ago

I had the same error while installing the package via conda (https://anaconda.org/conda-forge/explainerdashboard). It seems to be the case that the conda version is not the latest. Either you can downgrade your dtreeviz package or update your explainerdashboard version. At least for me, it worked after updating the package to the latest version (0.4.3).

oegedijk commented 11 months ago

So the conda-forge version was behind for a while because the CI could not deal with the numpy<1.25 requirement (that was needed due to numba and the latest numpy not playing nice together). It seems it works again with the latest numpy, so I removed the requirement and now the conda-forge build works as well, so now both PyPI and conda-forge are on 0.4.5 which should be compatible with the latest version of dtreeviz...