Riga Public Bus Transport System Analysis
During the hackathon in Riga OTN team (Pavel Hájek, Dmitrij Kožuch, Agris Šnepsts, Frantisek Kolovsk) was working on the analysis of public bus transport system. The data for such analysis was provided by the Riga City Hall (https://opendata.riga.lv/satiksme.html).
The datasets from Riga Open Data portal that were used by the team included
- Bus run timetable (Autobusu kustības saraksts)
- Electronic ticket validation data (E talonu reģistrācijas dati 09.02.2016). Because of the huge amount of data on a working day – more than 200 000 passengers validate their tickets – the data for just one working day (09.02.2016) were downloaded; the assumption was that on each working day the number of passenger trips will be similar
- Data for location and names of bus stops (Sabiedriskā transporta pieturas)
- Description of individual bus runs (Sabiedriskā transporta veidi)
The foregoing datasets are very good from the point of view that analysis of how much people use the bus routes can be conducted.
The team’s ultimate goal is to be able to evaluate public bus transportation system and eventually propose some optimizations (improvements). The reason behind is that it can be useful due to the fact that last year Riga public transport company required 93 mil. € subsidies. The analysis can help re-plan the bus routes to minimize subsidies.
First, some visualizations of where and how many people use public buses nowadays were made. For this it was necessary to do the mapping between validations of e-ticket and the bus schedules. In brief all rides where divided according to the bus route number, direction of the ride, start time (for validation interval it was the first validation on that certain bus ride) and end time (for validation interval it was the last validation on that certain bus ride). After this mapping was made – the cartographic visualizations were created and published as either WMS(-T) services or Map Compositions. At the moment they can be found at: http://opentnet.eu/web/guest/create-maps in Datasources (WMS(-T) services) and Compositions menus.
Some examples of such Map Compositions are provided below.
Figure 1. The number of passengers travelling on different bus lines and boarding buses at different stops and time of the day in Riga on 09.02.2016 (http://opentnet.eu/web/guest/create-maps?composition=261a62fd-d6f4-4f85-...)
Figure 2. Heatmap showing the number of passengers travelling on different bus lines and boarding buses at different stops and times of the day in Riga on 09.02.2016 (http://opentnet.eu/web/guest/create-maps?composition=6271e0ca-3d62-42c9-9644-40f1c46fc54d&hs_panel=layermanager)
As well as looking at overall picture and distributions of the passengers (total number of passengers at each stop by hour, total number of passengers at each bus route by hour etc.), the distribution of the passengers within each bus route but also in different particular parts of the city was also visualized. For this task WebGLayer.js library (https://github.com/jezekjan/webglayer ) was used. Below you can see the illustrations from the application.
Figure 3. Overall distribution of passengers on bus route #3 in Riga on 09.02.2016
Figure 4. The distribution of passengers on bus route #3 in Riga on 09.02.2016 from 5:00-7:00 where number of passengers was higher than 40
Figure 5. The distribution of passengers on bus route #3 in Riga on 09.02.2016 from 5:00-7:00 where the number of passengers was higher than 180
Figure 6. Comparing the distribution of passengers on buses in two randomly selected regions in Riga on 09.02.2016
For the WebGLayer application now more attributes are being prepared (the direction of the ride, land use, population density etc.) . The all in some time will be added to our application. And it will be possible to filter by all of them.
As we need very much to find some optimizations the basic evaluation parameters of the public bus transport network need to be assessed. Those are for instance: distance to the closest bus stop, bus network density, identifying overlapping portions of the routes et cetera.
Figure 7. 100/200/500/1000/2000-meters buffer zones around bus stops (http://opentnet.eu/web/guest/create-maps?composition=f910b59b-4e83-4b15-83c7-f2bf8b29e275&hs_panel=layermanager)
After looking at the current number and distribution of the passengers and evaluating the bus network, optimizations can be proposed.