Calculate Precision, Recall, and F1

You are given a list of binary classification outputs such as `[{"actual": 1, "predicted": 0, "confidence": 0.93}, ...]`, where class `1` is the positive class. Using the provided `actual` and `predicted` labels, write Python or pseudocode to compute the confusion-matrix counts (`TP`, `FP`, `TN`, `FN`) and then calculate: - `precision = TP / (TP + FP)` - `recall = TP / (TP + FN)` - `F1 = 2 * precision * recall / (precision + recall)` If the interviewer instead asks you to derive `predicted` from `confidence`, assume a decision threshold and explain how the metrics change when the threshold changes. Also describe how you would handle edge cases such as no predicted positives or no actual positives.