Skip to content

OpenDataDayZurich2016/visualization_delays

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

79 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Visualization of Delays in Verkehrsbetriebe Zürich (VBZ) Data

This repository collects code that was begun at ODD 2017 Zürich and has undergone subsequent elaboration and development.

It is connected to:

There are two ways that delays can accumulate during the course of a vehicle’s trip on a route. A delay can occur between stops, or a delay can occur at a stop. As a hack-a-thon activity we plotted for Route 33 the frequency of cases where the scheduled time between stops and the scheduled time at stops was exceeded by more than 2 minutes. We found that the vast majority of cases where schedules were exceeded was along the route, between stops.

We used QGIS to create a graphic that displays the relative counts of delays at stops and along route segments for Route 33. The two directions are shown by the split down the middle of each segment, with width proportional to the number of delays, regardless of their duration. This was the product for our work during during the hack-a-thon itself.

QGIS graphic

Subsequently we continued to expand the analysis. We set a minimum delay of one minute and examined the frequencies of delays of different lengths, as they arose at both stops and along segments. For Route 33, which is a trolly bus, it was not surprising that delays along route segments were relatively more frequent than delays at stops, and that this trend was reversed on Sundays, when there is likely much less vehicular traffic in the city.

Delays at stops and along segments

We then created scripts and functions to visualise the stops and associated delays with respect to the number of delays greater than an arbitrary threshold, over the course of 52 weeks, the week of 20.09.2015 to the week of 11.09.2016. These were output as Leaflet javascript map widgets.

The map widgets have mouse-over labels that indicate the name of the tram or bus stop and the number of delays that occurred over the 52 weeks. Currently, data for only one direction are shown, and with a minimum delay threshold of 1 minute. These are shown below, and in a separate report that was generated from the RStudio file.

The report on the html map widget exercise

—- Peter B. Pearman, Thomas Roth, and the Delay Visualization Team

Interactive Maps of Delay Counts at Stops for Selected Lines

Line 2

Line 7

Line 10

Line 14

Line 33

Line 72

Additional lines or delays in the opposite direction could easily be visualised with simple code modification.

===============================================================

Modifications

20.03.2017 adjusted_delays_further.Rmd provides code to go through 52 weeks of data and extract the stops that are relevant for each line. It also creates a bunch of empty .csv files for filling with delay data. text version of readme now provided to ease updating by other contributors

23.03.2017 The notebook file adjusted_stop_delays.Rmd tallies the number of delays greater than each of a series of six threshold values in minutes. It puts these values in a data set specific to a single bus or tram line. One can specify the values of the thresholds and the number of weeks of delay data for which one wants to extract the number of delays. It requires that a file exists for each line, as created by adjusted_delays_further.Rmd

06.04.2017 The file map.nb.html contains a notebook with code to make html widgets for lines that are represented in datasets that were created by the file adjustedstopdelays.nb.html

07.04.2017 Upload of html files, zipped html map widgets, map.Rmd source file.

09.04.2017 Upload of widgets, README.md

25.04.2017 Update widgets, code and report to reflect fix of calculation of delay, to exclude elapsed time at stop that occurs if the vehicle arrives early at the stop.

1.05.2017 Correct the README.md graphic of delay counts for Route 33. Corrected adjusted_delays_func.Rmd so that delays for Route 33 stops and along segments are both positive values.

2.05.2017 Update readme files and add QGIS PNG files.

3.05.2017 Update ready files, add scale bar to map widgets and replace all

4.05.2017 Add R User Group ‘Hackathon Flashback’ ppt presentation

Files with .np.html can be imported into RStudio for further work

The raw delay data are available here

==============================================================

Code guide

adjusted_delays_func.nb.html executes an initial analysis to compare the frequency of delays at stops and along segments between stops. This file is messy because it was used for many small, ad hoc inquiries regarding data characteristics.

adjusted_delays_further.nb.html executes additional code to construct .csv files for each route. This executes slowly because of the sapply function to remove delay time due to early arrival of vehicles. This could be removed (or go to an earlier commit) because there were very few early arrivals.

adjusted_stop_delays.nb.html tallies the number of delays during 52 weeks of data, for thresholds of 1.5, 2, 3, 4, 5, and 6 minutes or more. It also extracts all stops from the available data files from three years, and joins coordinate data for each stop appropriately. Actual threshold in the code should be checked carefully because these have often been changed.

map.nb.html reads the files constructed by the previous script and creates Leaflet html map widgets for several selected tram and bus lines.

.Rmd files can be used if rendered versions are not required.

Questions? Contact Peter Pearman [email protected]

html_map_widgets_and_report.zip has obvious contents

About

code of working group to visualize delays

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages