nearmiss — where the danger actually is

Road-hazard and near-miss rates, normalized by exposure, each with a 95% confidence interval and an n. Raw counts are report volume, not danger. The two maps below show the difference; the data table carries every finding and is the authoritative, non-visual equivalent of the maps.

⚠️ Showing the Davis synthetic demo dataset — generated test data, not real reports. The method is the point: swap in real reports with no code change.

Two maps, the same reports

The same near-miss reports, mapped two ways on a real street map. On the left, the raw report count — what most safety maps show: the busiest street looks the most dangerous. On the right, the rate per 1000 units of exposure — which is what actually reflects danger. Watch the busiest street recede and the real hotspot emerge. Nothing is conveyed by color alone: line thickness scales with the value, and significant hotspots are dashed and labeled. Everything here is in the data table, which the maps supplement.

① Raw report count — what most maps show
Loading map…
② Exposure-normalized rate — where the danger actually is
Loading map…

Ranked segments

Sort the table with the column buttons. Significance and confidence are stated in words, not by color.

Loading…
95% CI Confidence Quality flags
Loading published data…

How to read this