Considering the different output formats of models for different tasks, the framework implements different evaluators for different tasks and supports a variety of mainstream evaluation methods. The evaluation methods supported by different tasks are given below.
Task Name | Supported Metrics |
---|---|
Trajectory Location prediction |
|
Traffic Flow Prediction |
|
Traffic Speed Prediction |
|
On-Demand Service predition |
|
Origin-destination Matrix Prediction |
|
Traffic Accidents Prediction |
|
Map Matching |
|
Estimated Time of Arrival |
|
evaluator/traffic_state_pred evaluator/traj_loc_pred evaluator/map_matching evaluator/eta