Evaluators: improved documentation and more metrics - Githubissues

vanderschaarlab / autoprognosis

A system for automating the design of predictive modeling pipelines tailored for clinical prognosis.

https://www.autoprognosis.vanderschaar-lab.com/

Apache License 2.0

119 stars 27 forks source link

Evaluators: improved documentation and more metrics #62

Closed bcebere closed 1 year ago

bcebere commented 1 year ago

Description

Improve docs

Add multiple metrics for the classifier evaluation

[
 "aucroc",
 "aucprc",
 "accuracy",
 "f1_score_micro",
 "f1_score_macro",
 "f1_score_weighted",
 "kappa",
 "precision_micro",
 "precision_macro",
 "precision_weighted",
 "recall_micro",
 "recall_macro",
 "recall_weighted",
 "mcc",
]

Add "mse" and "mae" metrics for regression.

fixes https://github.com/vanderschaarlab/autoprognosis/issues/59 fixes https://github.com/vanderschaarlab/autoprognosis/issues/61

Checklist

[x] I have followed the Contribution Guidelines and Code of Conduct
[x] I have commented my code following the van der Schaar Lab Styleguide
[x] I have labeled this PR with the relevant Type labels
[x] My changes are covered by tests