[scikit-learn] Max f1 score for soft classifier?

[Sebastian Raschka]

>> Does scikit have a function to find the maximum f1 score (and decision
>> threshold) for a (soft) classifier?


Hm, I don't think so. F1-score is typically used as evaluation metric; hence, it's something optimized via hyperparameter tuning. There's an interesting publication though, where the authors modified the F1 score so that it's differentiable and can be used as a cost function for optimization/training: Maximum F1-Score Discriminative Training Criterion for Automatic Mispronunciation Detection: http://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=7055841

Best,
Sebastian

> On Jul 17, 2017, at 4:12 PM, Stuart Reynolds <stuart at stuartreynolds.net> wrote:
> 
> And... with that in mind -- are there methods that explicitly try to
> optimize the f1 score?
> 
> On Mon, Jul 17, 2017 at 9:41 AM, Stuart Reynolds
> <stuart at stuartreynolds.net> wrote:
>> Does scikit have a function to find the maximum f1 score (and decision
>> threshold) for a (soft) classifier?
>> 
>> - Stuart
> _______________________________________________
> scikit-learn mailing list
> scikit-learn at python.org
> https://mail.python.org/mailman/listinfo/scikit-learn