With 10-fold (outer) cross-validation:
- 70% accuracy
- 77% ROC AUC
With 2 lines of ML code, and interpretation (coefficients) that matches intuitions from provided Data Dictionary.
...
# the greatest single line of ML code, ever
lrcv = LogisticRegressionCV(cv=inner_cv, max_iter=max_iter, random_state=random_state)
results = cross_validate(lrcv, X, y, cv=outer_cv, scoring=scoring, n_jobs=n_jobs,
return_estimator=True, verbose=verbose)
...Interpretation from regression (see Jupyter notebook for more):
community.fico.com/s/explainable-machine-learning-challenge
