Simulation Results: A Closer Look I.e., Analysis, Validation, Visualization
Jan 16, 2016
Simulation Results: A Closer Look
I.e., Analysis, Validation, Visualization
2Learning Objectives
• Creating and analyzing traffic count comparisons
• Getting a rough idea of further analysis measuresand validation problems
• Using the On-The-Fly-Visualizer (OTFVis)
3Now!
• Start download of
– wuIST.mvi and– wuWU.mvi (> 60 MB, each)
– (zurich-switzerland.xml)
• @ http://www.matsim.org/node/412 → Exercises
→ Exercise 1 and (2)
4Traffic Counts
• Average hourly volumes of avg. working day• Future: Median, Variance, Max, Min (n days)
5Traffic Counts: Formats - KML
6Traffic Counts: Formats –TXT, HTML
7Traffic Counts: Parameters
Config file
<module name="…">…</module>
<module name="counts"> mandatory: <param name="inputCountsFile"
value="/path/to/counts.xml" /> <param name="outputformat" value="txt, html, kml" />
optional: <param name="countsScaleFactor" value="Double > 0 " /><param name="distanceFilter" value="Double >= 0" /> <param name="distanceFilterCenterNode"
value="String" /> </module>
8Traffic Counts: Parameters<param name="distanceFilter" value= "15000" /> <param name="distanceFilterCenterNode" value= "2531" />
Swisstopo
9Traffic Counts: Ex Post Creation
• Restraining the area of analysis
• Take
playground.anhorni.kti.CountsAnalyser or
playground.dgrether.analysis.CountsAnalyser
as an example
(will be implemented in the core soon)
• linkstats.txt → counts
10Traffic Counts Creation: Example CH
Raw count dataASTRA, cantons, cities
FilteringDIDOPublic HolidaysSummer HolidysOutliers
Averagingper station
Map MappingStations → LinksNAVTEQTELEATLASARE…
xml schemacounts_v1.xsdhttp://matsim.org/files/dtd/counts_v1.xsd
counts.xml
Parser/Writer
ConversionTXT → XML
11Further Analysis / Validation Measures
• Swiss Microcensus– Trip length and duration distributions– Modal Split
Spiders
Volumes Route switchers
12Validation Issues
• Lack of Validation Data and System Specification– Level of aggregation (analysis and modeling)
(e.g. count data vs. avg. trip length, route switcher analysis etc.)– MS = Sampling tool → Ensemble runs → confidence intervals
etc. (computational costs!)
• Future Measures– Travel speeds (underestimation: junction dynamics, non-
linearities?)– Facility loads (e.g., retailer consumer cards)
• A lot of interesting future work!
13Visualizers
OTFVisJava + OpenGL
MATVis ReplayC++
14OTFVis: Live Mode
config
Interactive queries in the live mode
15OTFVis: Playback Mode
mvi file
events file +
network file
16OTFVis: Playback Mode
Queries in the playback mode
Written in C++
Events not compatible mit Java Mobsim
MATVis Replay
(Daumenkinos)
Interactive queries in the live modelarge scale simulations: time and memory consuming
Room for participation
17Exercises – OTFVis: Westumfahrung
1. WU: Bypass Zurich
→ M. Balmer (15:30)
18Exercises - OTFVis
1. Building the bypass (WU) has made worse a prominent bottleneck in the region of Zurich. Try to locate it by using the OTFVis. It is best visible between ~ 7:15 and 9:00.
It is not visible very clearily in our results, but …
Hint:
19Exercises – OTFVis
Comparing 2 WU (Westumfahrung) scenarios
Scenario 1: actual state (base case)
=> convert events.run365.it150.txt.gz → wuIST.mvi 60
Scenario 2: with bypass (Westumfahrung)
=> convert events.run370.it220.txt.gz → wuWU.mvi 60
20Exercises - OTFVis
Comparing 2 scenarios with OTFVis
21Exercises – Google Earth
2. a) Using Google Earth with 100.countscompare.kmz look at the count stations in the south of Switzerland (Tessin, e.g., Lugano, Mendrisio).
b) Look at the count stations in the center of Zurich (e.g.,Rosengartenstrasse, Nordring)
c) What is the difference? What is the problem in question a)
22Exercises – Counts Configuration & OTFVis
Config file<module
name="counts"> <param name=„…
</module>
3. Configure a ZH simulation based on zurich-switzerland.xml such that counts are plotted only within a restricted area:
center = Bellevue and radius = 10 km.
I.e.; inspect network by OTFVis and define distanceFilterCenterNode in the count section of the config file.
23Exercises – „Solution 1“
e.g. 07:13
base case w/ bypass
24Exercises – „Solution 1“
base case w/ bypass
e.g. 08:27
25Exercises – „Solution“ 2
2. a) e.g., Mendrisio
26Exercises – „Solution“ 2
2. b) e.g., Rosengartenstrasse
27Exercises – „Solution“ 2
2. c) Boundary of modeling area
→ Task nbr 3!
28Exercises – „Solution“ 3
network.xml<link id="95214" from="156384139" to="300386714" … />
<link id="95215" from="300386714" to="156384136" … />
<param name="distanceFilter" value= "10000" />
<param name="distanceFilterCenterNode" value= "300386714" />