KDD2017_CUHK

KDD 2017 team from CUHK

Preprocessed Data Description

https://github.com/studdykid/KDD2017_CUHK/wiki/About-joined-csv-table

Data Feature:

Basic Info

Chinese Holidays:

There were two Chinese holidays during the time of the dataset.

Sep.15 to Sep. 17, 2016 was the Moon Festival holiday, while Sep. 18 (Sunday) was a working day.
Oct. 1 to Oct. 7, 2016 was the Chinese national holiday week, while Oct. 8 - 9 (Saturday & Sunday) were a working days.

During Chinese national holiday week, highway tolls were waived for passenger vehicles.

Road Network id

Road Network lanes

Road Network length

Checkpoints (Volume)

5 Checkpoint: 1-0 , 1-1, 2-0, 3-0, 3-1 2nd index{ 0: entry, 1: exit }

Lanes (Traj)

6 Paths: A-2, A-3, B-1, B-3, C-1, C-3

Datasets:

[Phase 1]
- Trajectories [2016, 7,19] - [2016,10,17]
- Volume       [2016, 9,19] - [2016,10,17]
- Weather      [2016, 7, 1] - [2016,10,17]

[Test Phase 1]
- Weather      [2016,10,18] - [2016,10,24]

[Phase 2]
- Trajectories [2016,10,18] - [2016,10,24]
- Volume       [2016,10,18] - [2016,10,24]
- Weather      [2016,10,25] - [2016,10,31]

[Test (Final)]
- Trajectories [2016,10,25] - [2016,10,31] (6:00-7:59 & 15:00-16:59 only)
- Volume       [2016,10,25] - [2016,10,31] (6:00-7:59 & 15:00-16:59 only)


=== In conclusion, totally we have... ===
[Training Dataset]
- Trajectories [2016, 7,19] - [2016,10,24]
- Volume       [2016, 9,19] - [2016,10,24]
- Weather      [2016, 7, 1] - [2016,10,24]

[Test (Final)]
- Trajectories [2016,10,25] - [2016,10,31] (6:00-7:59 & 15:00-16:59 only)
- Volume       [2016,10,25] - [2016,10,31] (6:00-7:59 & 15:00-16:59 only)
- Weather      [2016,10,25] - [2016,10,31] (whole day)

(All data above contain entire 24 hrs, except Final Test data.)

Features:

Volume (total) (20m)

ETC (20m)
Motorcycle (20m)
Cargocar (20m)
Privatecar (20m)
Unknowncar (20m)

routetime median (20m)

pressure (3 hrs)
sea_pressure (3 hrs)
wind_direction (3 hrs)
wind_speed (3 hrs)
temperature (3 hrs)
rel_humidity (3 hrs)
precipitation (3 hrs)

-- 15. date 16. hour 17. dayofweek 18. is_holiday

TODO:

Create the Input Tensor (3D)

Done:

Investigate Backfill for FillNA: Solved by interpolate (by Panda function)
Moving avg for 3hr's features (interpolate)

fortesit / kdd2017_cuhk_new Goto Github PK

kdd2017_cuhk_new's Introduction