A popular interview question for data scientists and technical program managers trying to get a job in the AI (Artificial Intelligence) field is to define the difference between precision and recall. In fact, one of my interviews at Facebook involved giving a definition of precision and recall. And while listening to interviews with data scientists, a popular questions is for the data scientist to list out all the classification metrics that he knows of such as accuracy, precision, recall, F1 score, sensitivity and specificity.

For newbies in the AI space, this requires a reasonable understanding of TP (True Positive), TN (True Negative), FP (False Positive) and FN (False Negative). The best way I found to explain these four possibilities is to start with a practical use-case involving counting people. And then to understand ghosts and skips!