Selective Columns — Solution & ExplanationProblems / Selective Columns / EditorialApproachThe table has 15 columns but we only need category and amount. The key optimization:

Project early — select only the 2 columns needed. In columnar storage like Parquet, this means 13 columns are never read from disk.

Aggregate — GROUP BY category with SUM, AVG, and COUNT.

Order — sort by total descending.

Why This MattersIn production data warehouses, tables often have 50-100+ columns. A SELECT * followed by aggregation forces the engine to read and process every column, even those not involved in the computation. Explicit column selection can reduce I/O by 90%+ in wide tables.Complexity

O(n) scan of only 2 columns instead of 15, with a single-pass aggregation.

Hints

You only need 'category' and 'amount' — don't SELECT *.

In PySpark, call .select('category', 'amount') before .groupBy().

Column projection in Parquet means unselected columns are never loaded.

Solve Selective Columns yourself →