SonoFlow nightly status
sodermalm · page built 2026-09-13 08:16 UTC
model is at
2026-09-12
5,550 nights absorbed
streets with a level
1,146
of 4,564 in the network
coverage, estimate
80.4%
nominal 80 %
coverage, forecast
71.4%
nominal 80 %
median error
0.105
forecast 0.197 (log)
change events
8
last began 2020-12-23
Coverage, last 3 runs
3 run(s) so far - the trend needs five before it means anything.
Last night
LAST NIGHT · sodermalm · 2026-09-12
=======================================
STATUS watch up
23 streets reported.
the network ran +32.1% against a normal night
the calendar expected +3.5% from holidays and season
so unexplained +27.6%
28% above what the calendar predicts.
change scores, one pair for the WHOLE network - never one street:
up 9.04 / 10 ##################
down 0.00 / 10
Past halfway and still climbing. Not confirmed; nothing to act on yet.
THE STREETS THAT REPORTED
------------------------
street vs normal calendar network own up down status
Söderledstunneln -19.2% +3.5% +89.4% -58.8% 0.00 9.79 watch down
Söderleden +27.0% +3.5% +37.3% -10.6% 2.74 0.00 ok
Johanneshovsbron +22.9% +3.5% +32.4% -10.4% 0.00 0.01 ok
Söderledstunneln +23.1% +3.5% +32.4% -10.2% 0.00 0.19 ok
Skansbropåfarten +27.8% +3.5% +37.3% -10.1% 0.00 0.00 ok
Skansbroavfarten +41.0% +3.5% +48.9% -8.5% 0.00 0.00 ok
Söderledstunneln +26.9% +3.5% +24.9% -1.8% 0.00 0.00 ok
Söderledstunneln +30.0% +3.5% +25.1% +0.3% 0.00 0.00 ok
Centralbron +30.1% +3.5% +25.1% +0.4% 0.00 0.00 ok
Söderledstunneln +30.2% +3.5% +25.1% +0.6% 0.00 0.00 ok
Söderledstunneln +31.4% +3.5% +25.6% +1.0% 0.00 0.00 ok
Söderledstunneln +32.7% +3.5% +25.6% +2.0% 0.00 0.00 ok
Johanneshovsbron +41.0% +3.5% +32.4% +2.8% 0.00 0.00 ok
Söderledstunneln +33.8% +3.5% +25.6% +2.9% 0.00 0.00 ok
Riddarfjärdsavfarten +48.6% +3.5% +39.4% +3.0% 0.00 0.00 ok
Riddarfjärdsavfarten +49.9% +3.5% +39.4% +3.9% 0.00 0.00 ok
Centralbron +35.3% +3.5% +25.6% +4.1% 0.00 0.00 ok
Södermalmsavfarten +33.5% +3.5% +21.9% +5.8% 0.00 0.00 ok
Riddarfjärdsavfarten +53.2% +3.5% +39.4% +6.2% 0.00 0.00 ok
Söderledstunneln +34.3% +3.5% +21.9% +6.4% 0.00 0.00 ok
Johanneshovsbron +42.3% +3.5% +24.9% +10.1% 0.00 0.00 ok
Riddarfjärdspåfarten +86.8% +3.5% +63.8% +10.2% 0.00 0.00 ok
Söderledstunneln +147.8% +3.5% +89.4% +26.3% 0.06 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 · sodermalm · as of 2026-09-12
====================================================
Of the times we published an 80 % interval, how often did the truth fall inside it?
STREETS WE MEASURE
estimate mode 80.4% honest
median error 0.105 log (x1.11)
forecast mode 71.3% TOO NARROW by 9 points
median error 0.197 log (x1.22)
over 5,550 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 80.4% 71.3% 5,550 -
last 365 days 88.8% 86.8% 172 63,932
last 90 days 89.9% 86.7% 90 50,612
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 48.5% TOO NARROW by 31 points
median error 0.704 log (x2.02)
over 15 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:
3357373856929621821 coverage 1.2% error x11.14 (168 cells)
6079980744524669011 coverage 22.4% error x12.03 (161 cells)
8954425914444133656 coverage 23.4% error x3.41 (4280 cells)
7831589162891272994 coverage 35.5% error x2.09 (791 cells)
7122657651002101828 coverage 40.7% error x1.69 (140 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
---------------------
24 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
1,146 of 4,564 streets (25.1 %) have ever been counted.
The other 3,418 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 1,146 counted, 46 were counted within 120 days and
carry the narrow measured-street band. The remaining 1,100 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 · sodermalm · as of 2026-09-12
================================================
8 confirmed changes. A change is reported when the network's day factor stays displaced for a run of dates, not when a single day looks unusual.
2010-02-04 UP +32.8%
began 2010-02-04 (when the shift started, not when it was noticed)
confirmed 2010-04-28 (144 processed dates in the run)
size +32.8% (+0.283 log) peak statistic 10.1
2010-08-19 UP +81.6%
began 2010-08-19 (when the shift started, not when it was noticed)
confirmed 2010-10-08 (115 processed dates in the run)
size +81.6% (+0.597 log) peak statistic 11.7
2012-05-24 UP +36.5%
began 2012-05-24 (when the shift started, not when it was noticed)
confirmed 2012-06-16 (27 processed dates in the run)
size +36.5% (+0.311 log) peak statistic 10.0
2013-08-01 UP +35.9%
began 2013-08-01 (when the shift started, not when it was noticed)
confirmed 2013-08-31 (33 processed dates in the run)
size +35.9% (+0.307 log) peak statistic 10.7
2013-11-21 UP +40.0%
began 2013-11-21 (when the shift started, not when it was noticed)
confirmed 2013-12-14 (26 processed dates in the run)
size +40.0% (+0.337 log) peak statistic 11.2
2014-07-12 DOWN -40.1%
began 2014-07-12 (when the shift started, not when it was noticed)
confirmed 2014-07-15 (35 processed dates in the run)
size -40.1% (-0.512 log) peak statistic 12.6
2020-03-15 DOWN -21.6%
began 2020-03-15 (when the shift started, not when it was noticed)
confirmed 2020-03-23 (75 processed dates in the run)
size -21.6% (-0.243 log) peak statistic 12.2
2020-12-23 DOWN -26.9%
began 2020-12-23 (when the shift started, not when it was noticed)
confirmed 2021-01-06 (15 processed dates in the run)
size -26.9% (-0.314 log) peak statistic 10.3
What this does NOT tell you:
· which streets. This watches the whole network's day factor, so it catches a shift
that moved everything. A single street changing is a different test (paper §8.3).
· why. Weather, roadworks, a new bridge and a policy change look identical here.
· that the model has corrected for it. A confirmed change should trigger a local reset;
until it does, forecasts for the affected streets stay wrong while every daily output
continues to look right (paper §8).
Each day: what the sensors said, and what the model published
Predicted flow across the whole network, hour by hour.
| street that reported | hours | measured veh/h | model was off by |
|---|
The model behind these numbers
Recent runs
| run | model at | nights | cov est | cov fc | mae est | changes |
|---|---|---|---|---|---|---|
| 2026-09-13 02:23 UTC | 2026-09-12 | 5,550 | 0.804 | 0.714 | 0.105 | 8 |
| 2026-09-13 01:22 UTC | 2026-09-11 | 5,549 | 0.804 | 0.713 | 0.105 | 8 |
| 2026-09-12 13:30 UTC | 2026-09-10 | 5,548 | 0.804 | 0.713 | 0.105 | 8 |