Precision at the indistinguishability threshold: a method for evaluating classification algorithms
This provides a better-motivated evaluation metric for classification tasks, addressing pitfalls in existing metrics like AUC and F1-score, though it is incremental as it builds on prior single-number metrics.
The paper tackles the problem of evaluating classification algorithms by proposing a new metric called 'precision at the indistinguishability threshold', which measures how often a randomly chosen image labeled as containing a cat actually contains a cat when the algorithm is tuned to make labeled and real cat images indistinguishable, with a 50% probability threshold.
There exist a wide range of single number metrics for assessing performance of classification algorithms, including AUC and the F1-score (Wikipedia lists 17 such metrics, with 27 different names). In this article, I propose a new metric to answer the following question: when an algorithm is tuned so that it can no longer distinguish labelled cats from real cats, how often does a randomly chosen image that has been labelled as containing a cat actually contain a cat? The steps to construct this metric are as follows. First, we set a threshold score such that when the algorithm is shown two randomly-chosen images -- one that has a score greater than the threshold (i.e. a picture labelled as containing a cat) and another from those pictures that really does contain a cat -- the probability that the image with the highest score is the one chosen from the set of real cat images is 50\%. At this decision threshold, the set of positively labelled images are indistinguishable from the set of images which are positive. Then, as a second step, we measure performance by asking how often a randomly chosen picture from those labelled as containing a cat actually contains a cat. This metric can be thought of as {\it precision at the indistinguishability threshold}. While this new metric doesn't address the tradeoff between precision and recall inherent to all such metrics, I do show why this method avoids pitfalls that can occur when using, for example AUC, and it is better motivated than, for example, the F1-score.