Create an F1 Score Metric for Binary Classification

[mfrasco]

Currently, the function f1 is used for information retrieval problems. However, I really want a function called f1_score that can be used for binary classification problems. This function should take two arguments, predicted and actual and compute the f1 score by calculating the point-wise precision and recall of the predictions.

[adam-m-mcelhinney]

@mfrasco, I would propose an fbeta_score function that accepts an optional beta arugment with a default set to one.

http://scikit-learn.org/stable/modules/generated/sklearn.metrics.fbeta_score.html
https://en.wikipedia.org/wiki/F1_score