Z-score normalise a spend array with NumPy

Question

Before clustering customers, monthly spend has to be put on a comparable scale. The standard move is a z-score: subtract the mean and divide by the standard deviation.

Write a function `zscore(values)` that takes a list of numbers and returns a **list** of z-scores, each rounded to 2 decimals. If t

Accepted Answer

Step 1: `np.asarray(values, dtype=float)` converts the list once and forces float maths, avoiding integer-division surprises. Step 2: Compute the standard deviation once and store it. NumPy's `.std()` is the *population* standard deviation (`ddof=0`); pass `ddof=1` if an interviewer asks for the sample version. Step 3: Guard against `sd == 0` — a constant array would otherwise produce `nan` from a divide-by-zero. Step 4: `(arr - arr.mean()) / sd` is a vectorised operation: NumPy broadcasts the scalar across every element, no Python loop required. Step 5: `np.round(..., 2).tolist()` converts back to plain Python floats so the result is JSON-friendly.

Z-score normalise a spend array with NumPy

More NumPy questions

Asked at Amazon