Table 1 Detailed information of the datasets.

From: Temporal representation learning enhanced dynamic adversarial graph convolutional network for traffic flow prediction

Datasets

Nodes

Edges

Time steps

Time interval (min)

Data type

Missingness (%)

Time range

METR-LA

207

1515

34,22

5

S

8.11

03/2012–06/2012

PEMS-BAY

325

2369

52,116

5

S

0.003

01/2017–05/2017

PeMS03

358

547

26,208

5

F

0.67

09/2018–11/2018

PeMS04

307

340

16,992

5

F, S, O

3.18

01/2018–02/2018

PeMS07

883

866

28,224

5

F

0.45

05/2017–08/2017

PeMS08

170

259

17,856

5

F, S, O

0.69

07/2016–08/2016

  1. In the “data type” column, ‘F’ represents traffic flow, ‘S’ represents traffic speed, and ‘O’ represents traffic occupancy.
Back to article page