Flag transaction outliers with the IQR rule

Question

Risk wants transactions flagged when they fall outside the classic 1.5 x IQR fence.

Write `flag_outliers(values)` that takes a list of numbers and returns a list of the values that are below `Q1 - 1.5*IQR` or above `Q3 + 1.5*IQR`, in their original order.

Accepted Answer

Step 1: `np.percentile(arr, [25, 75])` returns both quartiles in one pass, which is cheaper and less error-prone than two calls. Step 2: The interquartile range `q3 - q1` measures the spread of the middle half, so it is not dragged around by the extreme values you are hunting for. Step 3: The 1.5 multiplier is the conventional Tukey fence — quote it in interviews as a convention, not a law of nature. Step 4: Filtering the original list (not the array) preserves the input types and order the caller expects.

Flag transaction outliers with the IQR rule

More Statistics questions