Useful traffic information. Clear limits.
Dar Traffic shows on-demand estimates for 21 corridors and keeps independent history separate.
A little more context for the roads we share.
I started Dar Traffic because I wanted useful, honest information about the roads we use in Dar es Salaam. You can inspect a corridor, understand its estimate and see where the data comes from.
My longer-term goal is to train a useful traffic model. First I need independent data I have permission to use, and evidence that the model works. That work is still ahead.
From a request to a useful reading.
How the data reaches youCollect
An on-demand Google Cloud service requests traffic-aware estimates for configured endpoints. It does not store the response.
Validate
Each record is checked before display. Missing weather stays unknown and never inflates traffic.
Show the context
The latest timestamp, freshness, baseline and sample coverage stay visible alongside the estimate.
What does “extra delay” mean?
Extra delay is the traffic estimate minus the provider’s baseline, with a minimum of zero. The collector calculates this from seconds before rounding durations to minutes.
A zero estimate doesn’t prove that a road is empty. Different corridors can overlap, and their delays cannot be added to measure everyone’s time lost across the city.
Why no city efficiency score?
A defensible city-wide score would need coverage, vehicle counts and a validated method that we don’t currently have. We show the underlying estimates instead.
Sources and collection
Travel estimates: Google Distance Matrix. Weather context: Open-Meteo. Live service: Google Cloud Run. The website reads stored observations; refreshing does not start collection.
View the project and architectureForecasts need more than a label.
No validated prediction model is published. Google Maps content also has restrictions on model training. A future model needs permitted data and an evaluation against later observations.
Google Maps terms