paulj1989 / bulgarian-constitutional-court-decisions Goto Github PK

Describe the bug
I used the AUC-ROC curves from the yellowbrick package for testing model performance for both the logistic regression and the naive Bayes classifier, however, the same code doesn't work for the support vector classifier.

Instead, it produces the following error:

ValueError: Found input variables with inconsistent numbers of samples: [298, 149]

The code that produces the error:

roc_auc(svc, X_train, y_train, X_test=X_test, y_test=y_test, classes=[0, 1])

To Reproduce
Steps to reproduce the behavior:

1. Open baseline_models.ipynb
2. Run all cells in the baseline notebook
3. Scroll down to the SVC section of the notebook
4. Replace plot_roc_curve() with roc_auc(), using either the code from the previous model in the notebook, or the code above
5. Scroll down to see error

Expected behavior
The same AUC-ROC curve as shown under the first two models should be plotted.

Additional context
The plot_roc_curve() code, from sklearn, works as expected. However, it is less detailed than the yellowbrick plot (plus consistency would be preferred).

Is your feature request related to a problem? Please describe.
Model performance could be easier to find in the notebook, and easier to understand.

Describe the solution you'd like
Adding model performance and training visualizations would help illustrate the process.

Describe alternatives you've considered
N/A

Additional context
Some ideas for visualizing model performance can be found here and here.