🤖 AI drafts generated 25 Aug to 7 Sep 2026 · not reviewed · not recommendations
What this is
What was collected, and when
What the data shows
Written from the numbers in this archive. All times are UAE time (GMT+4). Filter by theme:
How long does each drive take at rush hour, compared with quiet hours?
Each row is Waze's best available drive-time estimate between two fixed points, not a measured trip, and not a named road (Waze picked the road). Quiet-hour = the fastest 10% of that route's own readings. Rush hour = the busiest of the 24 clock hours for that route, Monday to Thursday only (at least 100 readings on 20 days). It is not always the morning or evening hour you would expect. Speed = each reading's own distance divided by its own minutes, then the median, because Waze changes the route and so the distance. Minutes, UAE time. Click a row for its day. Ranked by extra minutes.
Which days are busy? Rush-hour drive time by day of the week
Average minutes in each route's rush hours (hours shown), for each day of the week, with the number of days behind each figure. UAE time. Shown for two Sharjah-border routes and, for comparison, the Marina and Downtown routes.
How bad is a bad day?
Rush-hour drive time (busiest hour, Monday to Thursday) on a typical day and on the worst 1 trip in 20, against quiet hours. Minutes. Ten routes with the biggest bad-day gap.
Route
Quiet
Typical rush
Bad day (1 in 20)
Best time to leave
For each route, the cheapest clock hour in the morning and evening windows (Monday to Thursday, UAE time) and the minutes saved against the worst hour in that window.
Route
Morning best
Evening best
Saves
When does the jam map show the most traffic? Hour of day by weekday
Jam map over time
Crash reports on the Waze map, by hour of day
Where the jam map showed slow or blocked roads
Waze live-map jam records, Dubai and the northern emirates. Last recorded jams = the last good sweep, levels 1 to 4. Recurring slow spots = places that jammed often across the whole study. Blocked or closed = level 5 segments (speed 0, about 130 m long) from the last sweep, plus cells that looked jammed at almost every hour of every day (roadworks, closures or staging areas); these are not counted as congestion.
Incidents in the Dubai Police open dataset
Records by type
Number of records. Types are my own grouping of the dataset's Arabic labels. Breakdowns and parking are not crashes.
Records by hour of day
Records per day
Records per day, and how many carry the dataset's own "serious" label. Days with no records are gaps in the dataset, not quiet days. Part of the early history was seeded and part was back-filled.
Incidents on a chosen day
Pick a day to see where the Dubai Police open dataset placed that day's records. Each dot is one record; red = the dataset's own "serious" label (a label, not an injury count). Locations are as published in the dataset and can be approximate. Days with no records are gaps in the dataset, not quiet days.
Hours 00:00–23:59
Incident types and the "serious" label
For each type in the Dubai Police dataset: how many records, what share carry the dataset's own "serious" label (a label, not an injury count), the hour it is most frequent, and how many sit near a traffic signal or roundabout (OpenStreetMap). This says nothing about fault or cause.
Flight arrivals at Dubai International (Dubai Airports open dataset)
Average arrival records per day at each hour of the day (UAE time).
Weather and drive times
Correlation (r) between how late each Dubai-to-Sharjah evening route was and dust (PM10), temperature, wind and visibility, after removing the usual time-of-day and day-of-week pattern.
Dust and the jam map
One dot per good jam-map sweep: dust (PM10, µg/m³) against jammed road (km, heavier jams count more; blocked segments left out).
Public holiday in the data
Date
Event
Does Waze ever suggest a side street first?
The pair is one trip (same start and end). One leg is the route Waze chose as its first choice (the main road); the other is a named side-street alternative. A side street can only appear as a lower-ranked alternative, so this table does not show what drivers actually do. It shows what Waze suggested.
City-wide side-street screen recurrence x timing x bypass
Every minor-street jam cluster in the city, scored against its nearest arterial bottleneck. Honesty rule: congestion observed is not through-traffic proven - candidates for probes, never conclusions.