causaltune
causaltune copied to clipboard
update requirements.txt to latest dowhy/econml releases
- The link to the dowhy package in requirements.txt is dead as the repo migrated from microsoft to pywhy
- upgrading dowhy to 0.8 seems to fix issue https://github.com/transferwise/auto-causality/issues/185
- note that dowhy 0.8 requires econml 0.13.1, so this would also need to be changed
Also update requirements.txt to hopefully get rid of specific version requirements for numpy and scikit-learn
econml 0.14 requires scitki-learn <1.1.3, otherwise get this error for IV models:
sklearn.utils._param_validation.InvalidParameterError: The 'cv' parameter of LogisticRegressionCV must be an int in the range [2, inf), an object implementing 'split' and 'get_n_splits', an iterable or None. Got <econml.sklearn_extensions.model_selection.WeightedStratifiedKFold>.
When there is a new econml release, test relaxing version requirement on scikit-learn