About joined csv table

Jump to bottom

ylchan87 edited this page Apr 8, 2017 · 2 revisions

About the csv table

It is made by exploreData/scripts/make_volume_table.py

Currently it has rows covering 09/20/2016 to 10/18/2016, at 3hr interval

About the columns

datetime : the time a row is for
dayofweek : 0=sun, 1=mon ... 6=sat
hour : hour of day
dt0_10_total_vol:
- dt0 means the info is valid for the same time as datetime
- 10 means tollgate=1, direction=0(entry to highway)
- total_vol means total num of car passing the tollgate
...
dt20_10_total_vol : similar to dt0_10_total_vol, but for 20min later
...
dt0_10_etc_vol : similar to dt0_10_total_vol, but for total num of car passing the tollgate with etc (electronic toll pay)
...
dt0_10_motorcycle_vol
dt0_10_cargocar_vol
dt0_10_privatecar_vol
dt0_10_unknowncar_vol: see "get_vehicle_class" func in make_volume_table.py for each car type's definition
...
dt0_A2_routetime_median : median of travel time from intersection A to gate 2, derived from trajectories_table_5
...
dt0_pressure
dt0_sea_pressure
...
dt180_pressure : weather data from weather table
...
is_holiday: 0-> not holiday, 1-> is a Chinese holiday

Too many columns?

To get started for predicting volume for tollgate 1, direction 0

we can just use

dt0_10_total_vol, dt20_10_total_vol, dt100_10_total_vol

to predict

dt120_10_total_vol, dt140_10_total_vol, dt220_10_total_vol

i.e. use first 2 hr volume to predict next 2 hr volume, as required by the task.

Clone this wiki locally