Datasets
This page defines datasets in ANYWAYS, their structure, behaviour, and the types currently supported.
Datasets are used to represent traffic assigned to a network and are combined within scenarios to simulate traffic.
What is a dataset?
Section titled “What is a dataset?”A dataset represents a set of trips with defined routing behaviour.
A dataset consists of:
- One or more origin-destination (OD) datasets
- An associated vehicle profile for each OD dataset
Each OD dataset defines:
- Where trips start and end
- How many trips occur
- For which vehicle type and time period
The associated vehicle profile defines:
- How those trips behave when routing through a network
A dataset does not contain:
- A network
- Fixed routes
- Traffic assignment results
Datasets are evaluated on a network within a scenario to calculate routes and traffic flows.
Dataset behaviour
Section titled “Dataset behaviour”When a dataset is evaluated on a network:
- Routes are calculated using the route planner
- Routing behaviour is determined by the associated vehicle profiles
- Traffic flows are derived per road segment and per vehicle type
The same dataset can produce different routing results when evaluated on different networks.
Dataset types
Section titled “Dataset types”ANYWAYS currently supports four dataset types.
Traffic counts dataset
Section titled “Traffic counts dataset”A traffic counts dataset is used to simulate existing traffic based on observed traffic counts.
This dataset type is defined by:
- A set of origin-destination (OD) pairs defined by the user
- A set of traffic counters placed on the network
- Observed vehicle counts per counter and vehicle type
- A set of possible vehicle profiles
Instead, ANYWAYS estimates traffic assignment by:
- Fitting an OD matrix for each selected vehicle profile
- Predicting routes through the network for each vehicle profile
- Optimising the fit between predicted traffic and observed traffic counts
The result is a set of OD matrix and vehicle profile pairs whose combined routing best matches the observed traffic counts. These fitted OD matrices and their associated vehicle profiles together form the resulting traffic assignment for the dataset.
The OD pairs can be defined in two ways: placed on the map by the user, or copied from another dataset in the same project. This also decides what the counters are taken to mean:
- Fitting — the counters are the traffic. Counted volumes are fitted onto the routes crossing them, and a route no counter sees carries nothing.
- Calibrating — the routes already describe the traffic and the counters set its level. Every route is scaled, including the ones no counter can see.
Fitting is the default for user-placed OD pairs, calibrating for copied ones; either can be chosen for either.
Modelled routes dataset
Section titled “Modelled routes dataset”A modelled routes dataset is built from the ANYWAYS model of everyday travel, by selecting an area on the map rather than by entering demand.
The model is a synthetic population: individuals with a home location and daily activities — work, education, personal business — each with a modelled departure hour and a route over the road network.
A modelled routes dataset is defined by:
- An area drawn on the map
- The vehicle mode of the trips to take
- The departure window, as whole hours of the day
- Which trips the area takes: those with their origin inside, their destination inside, and/or those passing through
- Whether trips are cut off at the area boundary
The trips are the model’s own: each carries the vehicle count and the routing profile the model assigned it, so no profile is selected for this dataset type. Trips cut off at the boundary have their origin and destination replaced by the points where the route enters and leaves the area.
Availability depends on coverage: the model holds routes for seeded regions only, and the dataset type is unavailable in a project whose area is not covered.
The resulting volumes are modelled, not observed, and the type is designed to be used together with a traffic counts dataset: the model supplies the origin-destination pairs, the counters supply the level they are scaled to. Copied into a traffic counts dataset set to calibrate, the two produce a baseline whose movements come from the model and whose volumes come from measurement — which is more than either type establishes alone.
On its own, a modelled routes dataset is a starting point where no counts exist.
Locations dataset
Section titled “Locations dataset”A locations dataset is used to simulate traffic travelling to or from a single location. This dataset type is commonly used for traffic impact assessments.
A locations dataset is defined by:
- A central location
- One or more distance bands (“donut” distances) around the location
- A number of randomly generated origin or destination locations within each band
Trips are generated:
- Towards the central location, or
- Away from the central location
Generated OD datasets are then assigned to the network using the selected vehicle profile.
Manual dataset
Section titled “Manual dataset”A manual dataset is used when users already have their own OD data or quickly want to test a very specific set of routes.
In a manual dataset:
- OD data is entered directly by the user
- The OD dataset is associated with a vehicle profile
This dataset type provides full control over demand definition and routing behaviour.
Relationship to other concepts
Section titled “Relationship to other concepts”- Datasets use networks as the basis for routing
- Datasets use vehicle profiles to determine routing behaviour
- Datasets are combined within scenarios by aggregation
- Routing is performed using the route planner
→ See Concepts
→ See Vehicle profiles
→ See Simulation behaviour
