Skip to content

Latest commit

 

History

History
25 lines (21 loc) · 1.45 KB

evaluator.rst

File metadata and controls

25 lines (21 loc) · 1.45 KB

Evaluator Introduction

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

TopK

Traffic Flow Prediction

MAE、MSE、RMSE、MAPE、R2、EVAR

Traffic Speed Prediction

MAE、MSE、RMSE、MAPE、R2、EVAR

On-Demand Service predition

MAE、MSE、RMSE、MAPE、R2、EVAR

Origin-destination Matrix Prediction

MAE、MSE、RMSE、MAPE、R2、EVAR

Traffic Accidents Prediction

MAE、MSE、RMSE、MAPE、R2、EVAR、Precision, Recall, F1-Score, MAP, PCC

Map Matching

RMF、AN、AL

Estimated Time of Arrival

MAE、MSE、RMSE、MAPE、R2、EVAR

evaluator/traffic_state_pred evaluator/traj_loc_pred evaluator/map_matching evaluator/eta