Fill missing product ratings the defensible way

Question

A ratings export has gaps. The analytics lead wants missing ratings filled with the **median rating of that product's category**, not a global constant.

Write `fill_ratings(df)` that returns a copy of the DataFrame with the `rating` column filled that way, rounded to 2 decimals. The original DataFr

Accepted Answer

Step 1: `df.copy()` keeps the caller's DataFrame untouched — mutating an input is the fastest way to create a bug nobody can reproduce. Step 2: `groupby('category')['rating'].transform('median')` returns a Series aligned to the original rows, so every row already knows its category median. Step 3: `fillna(medians)` replaces only the gaps; existing ratings are left exactly as they were. Step 4: `median` is chosen over `mean` because ratings are skewed by a handful of 1-star reviews.

Fill missing product ratings the defensible way

More Pandas — Cleaning questions

Asked at Amazon