Correction docstring

[autoih]

close #105

[hoffmansc]

The way I understood it, true labels in this context meant ground-truth.

so ...false_positives should read: weighted sum of predicted scores where true labels are 'unfavorable', since scores are always for the favorable label

and then ...false_negatives, e.g., should be weighted sum of 1 - predicted scores where true labels are 'favorable',

and finally, ...true_negatives should be weighted sum of 1 - predicted scores where true labels are 'unfavorable',

@nrkarthikeyan do you agree? Should we rephrase it as weighted sum of predicted 'scores' of instances with favorable ground-truth labels or something?

[autoih]

Thanks @hoffmansc. I've updated it.