Models
The right modelfor each kind of time series
Each model is built for one kind of time-series data. Pick a model to see what it does, what data it takes, and where it stands today.
Built for transportation: road-sensor speed, flow and other traffic series
Transportation time-series model · Preview
Segment by segment,reading the rhythm of a road network
SequenxAI Traffic is built for transportation time series — the speed and flow readings that road sensors record. It forecasts each road segment separately and attaches quantile intervals, so dispatch and operations teams can see how far things may swing. The model is in preview and not yet generally available.
Segment speeds · per-segment forecast Illustrative data
Illustrative data · 5-min sampling · 5 segments
The demo uses illustrative data, not live forecasts.P10–P90
Capabilities (planned)
What the model is being built to do
These capabilities are still being designed and validated. They become delivery commitments only after evaluation on a partner’s real data.
- 01
Per-segment forecasts
Forecast speed or flow for each road segment or sensor, not just one city-wide aggregate curve.
- 02
Quantile intervals
Return ranges such as P10–P90 instead of a single number, so teams can plan capacity and staffing buffers.
- 03
Evaluated against public benchmarks
Evaluation follows widely used public traffic datasets such as METR-LA, PEMS-BAY and EXPY-TKY; no comparison results have been published yet.
- 04
Data cleaning and quality reports (planned)
Sensor dropouts, clock drift and sudden jumps are cleaned first, with a quality report; the exact scope is agreed in writing.
Inputs · Outputs
The data the model works with
Inputs (planned)
- Speed, flow and similar time series per road segment or sensor
- Sampling interval depends on the setting, commonly 5 to 10 minutes
- Weather, holidays and event calendars as extra context (planned)
Outputs (planned)
- Median forecast with quantile intervals for each segment
- Uncertainty ranges you can compare across segments
- No signal timing or autonomous-driving control commands
Applications
Where the model is applied
Scope & boundaries
What the model does —and doesn’t do
It is:
- A forecasting model specialised for transportation time series
- A reference forecast for dispatch, capacity and staffing decisions
- In preview; delivery scope agreed with each customer per scenario
It is not:
- Generally available, or live in any city traffic system
- A signal-control or autonomous-driving system — it issues no control commands
- A replacement for the final decisions of traffic authorities and operators
- A model that claims accuracy leadership without same-scene comparison
Start an evaluation
Evaluate the modelon your traffic data
Start with the traffic demo, then agree an evaluation scope with us. Delivery is defined by written agreement.