Unpivot a wide survey export into tidy rows

Question

The survey tool exports one column per question. Analysis needs tidy long format instead.

Write `tidy(df)` that melts the `q1`, `q2` and `q3` columns into `question`/`score` pairs, keeps `respondent`, drops rows with a missing score, and returns the result sorted by respondent then question with a

Accepted Answer

Step 1: `melt` is the inverse of `pivot`: `id_vars` stay as identifying columns while `value_vars` collapse into two columns. Step 2: `var_name` and `value_name` name the output columns explicitly instead of accepting the default `variable`/`value`. Step 3: `dropna(subset=['score'])` removes unanswered questions — in wide format those were invisible `NaN` cells. Step 4: Sorting and resetting the index gives a stable frame that is safe to join or export.

Unpivot a wide survey export into tidy rows

More Pandas — Cleaning questions

Asked at Deloitte