Definitions
Each measure has a boundary and an aggregation limit. Ignoring them produces false trends.
| Measure | Definition and boundary | Limit on use |
|---|---|---|
| Passenger trains/day | Published daily service count, not train fleet; gauges and Metro scope matter | Do not equate to distinct train numbers or daily departures without table definition |
| Passenger speed | Published overall average including halts, by BG class | Do not average class speeds; use compatible distance/time totals if available. Timetable vs actual running logs must remain separate |
| Goods speed | BG all traction, excluding departmental, ASS20 / yearbook | Published aggregate until denominator method independently reconstructed. COVID mix is exceptional |
| Accident count | Historical report and modern consequential counts are separate regimes | No70-year homogeneous decline percentage; retain KRCL/UMLC and definition thresholds |
| Published accident index | Rounded accidents per million train-km | Do not infer exact exposure by dividing rounded index; count scope must match |
| Route / track length | Route-km, running-track-km and total-track-km differ | Stocks at31March; preserve gauge/AC/DC and network scope |
| Investment | Capital-at-charge, capital investment and total investment are nominal stocks | Not annual capex; no stock differencing as expenditure |
| Context rows | 574 source observations including benchmarks/vintages | QA is row-specific; rendered table does not verify every cell |
| ASS rows | Official national gauge/traction panel,2002-03..2024-25 | 235 raw table rows; not individual train records and not yet zone-level |
| FY | 1April to31March;1956-57..2025-26 is70FY spine | Earlier benchmarks and2026-27 partial remain outside completeFY comparison |