How to clean a CSV file safely

Keep the original file immutable, profile types and missingness, resolve structural problems first, apply narrowly scoped transformations, and verify row counts, null counts, unique IDs, and important totals after each step.

Profile and clean a CSV file without silent data loss: back up the source, inspect structure, fix one issue class at a time, and validate every change.