# Transformation log

Example project: the café, people and data are invented.

The file as supplied (`cafe-survey-original.csv`, 412 rows) is kept unchanged. Every change below is made by `analysis.py`, which writes the checked file `cafe-survey-checked.csv` (400 rows). Running the script again gives the same counts.

| # | Change | Rows | Rule |
|---|---|---|---|
| 1 | Removed exact duplicate submissions | 12 | Every field identical, including response ID and time. The first copy is kept. |
| 2 | Converted waiting time to whole minutes | 311 | Text such as "10 mins" or "about 6 min" becomes 10 or 6 in a new `wait_minutes` column. The first number in the answer is used. The original `wait_time` column is kept. |
| 3 | Left blank waiting times as missing | 18 | Not estimated or filled in. |
| 4 | Left blank ratings as missing | 30 | Not estimated or filled in. |
| 5 | Set ratings outside 1–5 to missing | 2 | Values 0 and 6 are not on the scale; the form allowed them by mistake. |
| 6 | Trimmed spaces around comments | 55 | Wording unchanged. |

Nothing else was changed, and no rows were removed except the duplicates.

After the checks: 400 responses, 368 with a usable rating, 350 with both a rating and a waiting time, 334 with a comment.