# ECRS process time exercise materials

BizThinkFrame created every record for this fictional teaching exercise. These are not company data. The simulated period is October 1–4, 2026. All 12 routine requests are complete; there are no incomplete, complex, or overnight requests. The records are not evidence of team service levels or method effectiveness.

## Files and reproduction

- `process-log.csv`: one request per row, with stable IDs, four timestamps with UTC+08:00 offsets, five active-effort categories, review queue time, and an approval-before-handoff flag.
- `process-summary.json`: script output containing definitions, the source CSV's SHA-256, totals, means, and a conditional scenario. It is not a measured improvement result.
- `analyze_process.py`: Python 3 standard library only. Checks IDs, timestamp order, stage durations, approval flags, and the serial-process assumption before calculating results.
- `process-worksheet.md` / `process-worksheet.en.md`: separate Chinese and English worksheets with one worked example. Use the English file for this article.

Download the CSV, script, and summary into one directory, then run:

```sh
python3 analyze_process.py
```

The output JSON should match the supplied summary. To save it, use a different file so that the reference summary remains available:

```sh
python3 analyze_process.py > recalculated.json
```

## Fields and calculation basis

`entry_minutes`, `copy_minutes`, `review_minutes`, `handoff_minutes`, and `rework_minutes` measure staff effort. Rework includes completing missing fields and checking them again after initial review; it is counted once in active effort. `wait_minutes` is elapsed queue time between `queued_at` and `review_started_at`. It does not measure staff inactivity or recoverable effort.

Within each request, work is sequential, only one person works at a time, and there is one wait before review. Rework has no additional wait. Receipt to queue entry equals entry plus copying; review start to handoff equals review plus rework plus notification. Lead time is independently calculated from `handoff_at - received_at` and checked against work plus waiting. Do not reuse that addition rule for concurrent work in a real process.

Expected results: 216 staff-minutes of work, 504 request-minutes of waiting, and 720 minutes of summed lead time, with means of 18, 42, and 60. Six requests needed rework, totaling 24 staff-minutes. Removing all copying with no added effort would avoid 36 staff-minutes for this batch and leave 180. This is a conditional estimate, not a statistical confidence bound or a promise for future batches.

## Use and evidence limits

Use these materials for learning or internal process exercises, retaining the BizThinkFrame credit and fictional-data label. With real data, redefine period, scope, concurrency, missing records, and incomplete requests; do not remove exceptions to create apparent improvement. The script checks this example's format and assumptions. It is not a general process diagnostic tool, and the framework page does not automatically import or calculate these files.

Method reference: JICA / Uzbekistan Japan Center, Advanced KAIZEN course, October 2016, slides 133–134, PDF page 211: [original materials](https://openjicareport.jica.go.jp/pdf/12334025_03.pdf#page=211), accessed October 6, 2026. The source supports the four improvement prompts, not the simulated counts, the 36-minute scenario, or pilot approval.