nacka · page built 2026-09-26 18:01 UTC

Coverage, last 23 runs

Forecast mode. The dashed line is the 80 % the interval claims; below it the published band is too narrow.

Last night

LAST NIGHT  ·  nacka  ·  2026-09-25
===================================

  STATUS   ok

      11 streets reported.

      the network ran           +0.2%  against a normal night
      the calendar expected     +0.2%  from holidays and season
      so unexplained            +0.0%

      The calendar accounts for the night. Nothing to explain.

      change scores, one pair for the WHOLE network - never one street:
        up     0.00 / 10  
        down   0.00 / 10  

      No alarm. 10.0 to go.


  THE STREETS THAT REPORTED
  ------------------------

  street                    vs normal  calendar  network      own      up  down  status
  Värmdöleden                   -6.0%     +0.2%    +0.0%    -6.2%    0.00  0.00  ok
  Värmdöleden                   -5.0%     +0.2%    +0.0%    -5.3%    0.00  0.00  ok
                                -3.4%     +0.2%    +0.0%    -3.7%    0.00  0.00  ok
  Värmdöleden                   +3.1%     +0.2%    +0.0%    +2.9%    0.00  0.00  ok
                                +4.1%     +0.2%    +0.0%    +3.9%    0.00  0.00  ok
  Sicklatunneln                 +5.1%     +0.2%    +0.0%    +4.9%    0.00  0.00  ok
  Värmdöleden                   +6.8%     +0.2%    +0.0%    +6.5%    0.00  0.00  ok
                                +8.5%     +0.2%    +0.0%    +8.3%    0.00  0.00  ok
  Värmdöleden                   +9.3%     +0.2%    +0.0%    +9.0%    0.00  0.00  ok
  Nackatunneln                 +15.0%     +0.2%    +0.0%   +14.8%    0.00  0.00  ok
  Värmdöleden                  +15.2%     +0.2%    +0.0%   +15.0%    0.00  0.00  ok

  'own' is what the street did beyond the calendar and beyond the whole network's night;
  the two totals accumulate it. One pair per street, not one for the city.

Honesty statement

HONESTY STATEMENT  ·  nacka  ·  as of 2026-09-25
================================================

Of the times we published an 80 % interval, how often did the truth fall inside it?

  STREETS WE MEASURE

      estimate mode    99.2%   too wide by 19 points
                      median error 0.055 log  (x1.06)
      forecast mode    99.2%   too wide by 19 points
                      median error 0.055 log  (x1.06)
      over 8 nights.  The gap between the two modes is what knowing the day
      is worth, measured rather than asserted.

      Lately, on the same streets, cell-weighted the same way:

        window          estimate  forecast   nights    cells
        all time           99.2%     99.2%        8        -
        last 365 days      99.2%     99.2%        8    2,088
        last 90 days       99.2%     99.2%        8    2,088

      CALENDAR days, not the last N scored nights: the sensor fleet thinned, so
      the last 365 SCORED nights span over three calendar years here and would
      be a misleading thing to call 'the last year'. A window with few nights is
      a weak signal, which is why the counts are beside the figures.

  STREETS WE DO NOT

      *** THIS NUMBER IS FROM AN OLDER RUN. validation_edges.parquet predates
          the model beside it, so the two describe different fits. Re-run with
          validation enabled - `pipeline <area> --until <date> --kfold` - before
          quoting it. ***
      coverage         61.5%   TOO NARROW by 19 points
                      median error 0.669 log  (x1.95)
      over 4 streets whose counts were hidden and then predicted exactly
      as an unmeasured street is. This is the claim the product rests on.

      Worst-covered, which the pooled figure hides:

        8428924605572734317   coverage 53.6%   error x2.11   (168 cells)
        9060005599005804645   coverage 58.9%   error x1.90   (168 cells)
        6143810180047670148   coverage 61.3%   error x2.00   (168 cells)
        5737474541971312101   coverage 72.0%   error x1.90   (168 cells)

      A pooled number blends a well-calibrated majority with a few streets that
      are confidently wrong. Both are worth reporting; only one is a tuning
      problem.


HOW MUCH IS REPORTING
---------------------

      11 streets report on a typical recent night (at or above the 5 the day factor needs).

  HOW MUCH OF THE NETWORK IS MEASURED AT ALL

      40 of 22,191 streets (0.2 %) have ever been counted.
      The other 22,151 are inferred from the network. A city where
      3 % of streets carry recent measurement is a different product from one where
      30 % do, and a uniform-looking map should not imply uniform confidence.

      Of the 40 counted, 40 were counted within 120 days and
      carry the narrow measured-street band. The remaining 0 are STALE -
      counted once, long ago - and are banded as inferred streets, because that
      is what they now are. Their level comes from the drift and their neighbours,
      not from their own sensor.

Change alerts

CHANGE ALERTS  ·  nacka  ·  as of 2026-09-25
============================================

No confirmed changes.

Each day: what the sensors said, and what the model published

Predicted flow across the whole network, hour by hour.

The model behind these numbers

Recent runs