BenfordSample report
Benford anomaly preview
This is a static sample built to preview Benford, Last Two Digits (LTD), and Benford Subset Divergence Analysis (BSDA) before connecting your own datasets.
First Digit (Sample)
Example metrics: sample=12,453 • MAD=0.007214 • SSD=8.34 · Acceptably Close (Kossovsky Fig. 3.3)
First Two Digits (Sample)
Example metrics: sample=12,453 • MAD=0.001921 • SSD=1.27 · ≈ Perfectly Benford (Kossovsky Fig. 3.3)
Last Two Digits (Sample)
LTD uses a flat 1% expected rate per bin (00–99), not Benford log curves—no confidence bands. Example: raw=12,453 • LTD-eligible=12,401 • SSD=142.8 · Non-Benford (Kossovsky Fig. 3.3)
Subset Divergence Analysis (Sample)
Pro Benford subset screening ranks slices within each dimension column—for example every expense category value, then every region. Multi-way combinations across dimensions (such as region × expense category) are planned for a future release. Click a row to preview the scoped first-two-digits chart.
Run settings (snapshot): first-two-digits SSD • min sample 2,500 • top 25 per dimension
Ranks category slices by first-two-digit SSD. Scanning many subsets increases the chance of statistically unusual but benign segments; SSD bands are not proof of fraud or error. Bands describe statistical distance from the expected distribution; they are not evidence of fraud or error.
Amount
Subset leaderboard by dimension column (first-two-digits SSD, descending). Click a row to view its first-two-digits chart.
| Rank | Slice | n | MAD | SSD | Top spikes | |
|---|---|---|---|---|---|---|
| 1 | Travel | 3,500 | 0.013479 | SSD=2574.4 • Non-Benford | 49 (+48.6 pp), 48 (+12.0 pp) | |
| 2 | General | 3,500 | 0.000412 | SSD=6.5 • Acceptably Close | 32 (+4.8 pp) |
Value repetition (Sample)
Repetition checks help surface values that appear unusually often.
- Value
repeated observations: 1,982 / 12,453 (15.92%)
repeated unique values: 218
top repeated: 99.99 (143), 1,000.00 (127), 49.95 (118), 250.00 (104), 19.99 (99)