Feature request: regression report

Describe the workflow you want to enable

I’d like an alternative to the wildely used sklearn.metrics.classification_report for regression problems

Describe your proposed solution

I purpose to implement it as sklearn.metrics.regression_report, following code is a draft I code my self to re-use:

from sklearn.metrics import mean_absolute_error, mean_squared_error, max_error, median_absolute_error, r2_score, explained_variance_score
import numpy as np

def regression_report(y_true, y_pred):
    
    error = y_true - y_pred
    percentil = [5,25,50,75,95]
    percentil_value = np.percentile(error, percentil)
    
    metrics = [
        ('mean absolute error', mean_absolute_error(y_true, y_pred)),
        ('median absolute error', median_absolute_error(y_true, y_pred)),
        ('mean squared error', mean_squared_error(y_true, y_pred)),
        ('max error', max_error(y_true, y_pred)),
        ('r2 score', r2_score(y_true, y_pred)),
        ('explained variance score', explained_variance_score(y_true, y_pred))
    ]
    
    print('Metrics for regression:')
    for metric_name, metric_value in metrics:
        print(f'{metric_name:>25s}: {metric_value: >20.3f}')
        
    print('\nPercentiles:')
    for p, pv in zip(percentil, percentil_value):
        print(f'{p: 25d}: {pv:>20.3f}')

Usage example:

>>> from sklearn.metrics import regression_report
>>> regression_report(y_train, y_train_pred)
>>> Metrics for regression:
      mean absolute error:               10.176
    median absolute error:                6.714
       mean squared error:              247.877
                max error:               82.595
                 r2 score:                0.565
 explained variance score:                0.574

Percentiles:
                        5:              -14.780
                       25:               -6.320
                       50:               -0.271
                       75:                7.300
                       95:               34.608

Describe alternatives you’ve considered, if relevant

Relevant metrics and inpout/output format should be discussed by the community according to its interest and usage, and keeping the interfaces as similar as possible to the related classification_report.

Additional context

I’m able to implement and raise a PR if there is interest on it. I’d love to contribute to this library, Just would like to know if there is interest for this and how/which metrics to include in the report

[nasserboan]

Hello @rola93! Nice addition, have you made any pull request with this code? I’m interested in helping with this implementation.

[Ratansingh648]

Hi @rola93 , @nasserboan I have already started implementing this feature today. Will send a pull request soon. Investigating if some more attributes that are used commonly can be added to this.

Thanks !