Loading page...Загружаем страницу...正在加载页面...
AtlasRepo

Demo data that has to raise an alarm, without faking the total

Description and use

Synthetic payroll data generated without manually setting totals now reliably triggers a 3% overspend alarm, adhering to department deviation rules and validation checks.

Problem

Generate synthetic payroll data that triggers a 3% overspend alarm without manually setting the total.

Outcome

Synthetic payroll data now consistently triggers the 3% overspend alarm while maintaining realistic department deviations and passing all validation checks.

Workflow

  • Define monthly deviation rule (1–3%) for each department.
  • Seed generator to produce consistent CSV.
  • Attempt 1: all departments within ±1–3% → +0.82% total.
  • Attempt 2: Engineering at bottom, others at top → +2.16% total.
  • Attempt 3: Allow Sales to deviate 9–10% → +3.25% total.
  • Add exception rule: only one department may leave the range.
  • Add checks: yearly alarm threshold, Q4 > Q1–Q3, single exception.
  • Re‑write check to use department trend + uplift instead of plan + uplift.
  • Validate with 11 checks, ensuring all pass.
  • Deploy dashboard as single HTML file (~680 KB).

Evidence

Each department gets a monthly deviation from plan of 1 to 3 percent, always in the same direction.

Source

Round 3 is tuning, and I would rather say so. I didn't set the total, but I chose which department breaks out knowing what it would do to the year.

Source

I rewrote that check and added three: the year must cross the alarm threshold, Sep-Dec must run hotter than Jan-Aug, and exactly one department may leave the range.

Source

  • Record date: 2026-10-01T10:55:44.056Z