This is the project page for ETHz Data Science Lab 2023. We collaborated with ETH IVT Group and completed a detailed analysis on the traffic sensor data of the Swiss federal highway, Zurich canton and Zurich city. Due to data privacy, we are not able to publish the original dataset but we are allowed to publish our analysis report. In this repo you can find our conclusions as well as the drawing utilities we draw our figures (easy-to-use plotly plugin functions). You can also find the code for generating the figures in report.ipynb but you are not able to run it without the data.
- Shah Akshat akshah@student.ethz.ch
- Jingyu Wang jingyuwang@student.ethz.ch
- Shakhnovich Kirill kirillsh@student.ethz.ch
Without any preprocesisng, we have three(four) data set:
Canton:
Canton data are saved in files in the format of 10 digits iddate, where id is the sensor id and date is of YYMMDD, eg. 1888210929. It starts with a frontmatter contains metadata
* BEGIN
* FORMAT = VBV-1
* FORMATTER = GRFORMAT Release = 3.5.47
* INSTRUMENT = M680 Serial = 705172 Release = 2.9
* FILENAME =
* SITE = 1888
* LOCATION = Moenchaltdorf
* GRIDREF =
* HEADINGS =
* STARTREC = 00:00 29/09/21
* STOPREC = 00:00 30/09/21
* BATTERY = 6.90 6.90
* SENSORS = LL LL NONE NONE NONE NONE NONE NONE
* DATEFORM = DD/MM/YY
* UNITS = Metric
* PRUNITS = KPH-CM-10KG
* CLASS = SWISS10
Then the recorded data:
* SITE HEAD DDMMYY HHMM SS HH RESCOD L D HEAD GAP SPD LENTH CS CH
1888 000001 290921 0000 03 00 000000 2 1 3.0 3.0 49 1851 1 L
1888 000002 290921 0000 35 00 000000 1 1 35.0 35.0 51 402 3 L
Each column means
SITE -- detector id
HEAD -- counter
DDMMYY -- day month year
HHMM -- time hour minutes
SS -- time seconds
HH -- time 1/100s
RESCOD -- na
L -- lane id (?)
D -- direction of travel (1= normal, 2= backward)
HEAD -- time gap (front bumper to front bumper)
GAP -- time gap (front bumper to back bumper)
SPD -- speed in km/h
LENTH -- vehicle length in cm
CS -- type of vehicle: 1= bus, 2= motorcycle, 3= car, 4= car with trailer, 5= delivery, 6= delivery with trailer, 7= delivery with long trailer, 8= truck, 9= long truck, 10= truck with trailer
CH -- height of a vehicle: very low, low, medium, high
Astra:
Astra data are saved in csv files in the format of id_date.csv, where id is the sensor id and the date is of YYYY-MM, eg. 20_2021-01.csv. It contains the following information
# Column Dtype Remark
--- ------ ----- ------
0 _id object event_id
1 _class object
2 src_time object event time
3 src_time_src int64
4 src_time_inc int64
5 rcv_delay int64
6 prot_type int64
7 zs_id int64 sensor_id
8 vd int64 lane_id (sensor_id+lane_id identifies one sensor)
9 vd_seqnum_type int64
10 vd_seqnum_val int64
11 vd_status int64
12 vd_class_type int64
13 vd_class_val int64 vehicle class
14 vd_length_type int64
15 vd_length_val int64 vehicle length
16 vd_dir object
17 vd_speed_type int64
18 vd_speed_val int64 vehicle speed
19 vd_occ_type int64
20 vd_occ_val int64 occupancy
21 vd_head_type int64
22 vd_head_val int64
23 vd_gap_type int64
24 vd_gap_val int64 City:
City data are stored per year in csv files in the format of Zurich_YYYY_raw2.csv
# Column Dtype
--- ------ -----
0 day object
1 interval int64
2 detid object
3 flow float64
4 occ float64
5 valid int64
6 suspect int64
Annotation
interval = seconds from midnight
flow = flow per hour in the interval
occ = occupancy in percentage of time during which the detector was occupied (i.e. covered by a vehicle)
valid= the system classified the measurement as a valid one (without errors)
suspect= the system classified the measurement as suspicious (maybe an error)
Example:
day interval detid flow occ valid suspect
0 2021-01-01 359 K126.D1 0.000000 0.000000 1 0
1 2021-01-01 539 K126.D1 2.941176 0.008987 1 0
2 2021-01-01 719 K126.D1 0.000000 0.000000 1 0
3 2021-01-01 899 K126.D1 0.000000 0.000000 1 0
4 2021-01-01 1079 K126.D1 0.000000 0.000000 1 0
And the corresponding GIS data:
Canton:
detid geometry
0 canton_88_1 POINT (8.60388 47.41007)
1 canton_88_2 POINT (8.60369 47.41000)
2 canton_103_1 POINT (8.61135 47.33826)
3 canton_103_2 POINT (8.61124 47.33830)
4 canton_108_1 POINT (8.44618 47.26777)
Astra:
detid geometry
0 astra_2_1 POINT (6.72462 46.57509)
1 astra_2_2 POINT (6.72462 46.57515)
2 astra_3_1 POINT (8.69818 46.10508)
3 astra_3_2 POINT (8.69823 46.10506)
4 astra_5_1 POINT (9.07607 47.00019)
City:
detid type geometry
0 K2.D11 IMP POINT (8.51846 47.37551)
1 K2.D13 IMP POINT (8.51823 47.37519)
2 K2.D14 IMP POINT (8.51875 47.37497)
3 K2.D15 IMP POINT (8.51880 47.37487)
4 K2.D16 IMP POINT (8.51884 47.37487)
Please find our poster below to see our main conclusions.
