Skip to main content
RISE logo

Master thesis: Road traffic analysis for Stockholm’s major road(s)

Summary
Like any European capital, Stockholm suffers from many problems related to its road network. The main one being traffic jams, which are aggravated with difficult weather conditions in winter, for example, but also when accidents occur. This work aims at predicting and understanding the behavior of this network based on data collected in several places, taking into account the different variables involved that influence the way people’s choice, and therefore, leading to a given state of the traffic network. More specifically, the goal is to predict and model the traffic flow, i.e macroscopic information, via ground measurements called Motorway Control System, using microscopic data called floating data, provided by INRIX, typically in car. One a prediction can be found, there is little need for complex, expensive, hard to maintain infrastructure equipment.

Procedure
Steps: Predict flow or density from macroscopic data (Mcs) from microscopic data (INRIX)

  • Literature review
  • Especially [Seo2019] and [Kim2014] and [Tsanakas2019-lic] (see below)

- Understand the basics of the road traffic (fundamental)

  • Look at the MCS (motorway control system) data

- Identify relevant features to be used
- Extraction and pre-processing
- Data exploration

  • Look at INRIX data

- Same steps as above

  • Compare the MCS and INRIX for speed correlations. Which traffic effects are relevant?

- Lanes?
- Time of day?

  • Data driven traffic prediction (i.e. domain specific)
  • Linear regression (“regress towards the mean”)
  • Linear

- https://en.wikipedia.org/wiki/Anscombe's_quartet
- Regress on Y ~ X or X ~ Y
- Locally weighted regression (LOESS)
- Is regression the correct method?
- E.g. Usual cases Height ~ Weight
- Our data is not in this form
- Flow / Density is a non-linear func. of speed (see 8)
- Use of a single INRIX variable?
- Or multiple (Speed + ?)

  • Generalised regression

- Non-linear

  • ML time series prediction

- Recurrent neural network approaches (RNN)
- Long short time memory (LSTM)

Terms
Scope: 30 hp.
Supervisor: Ian Marsh, Ph.D, +46707721536
Placement: Kista, Stockholm.

References
[Tsanakas2019-Lic]  Tsanakas, Nikolaos, Emission estimation based on traffic models and measurements, 2019, Licentiate thesis, monograph.

[Kim2014] Seoungbum Kim, Benjamin Coifman, Comparing INRIX speed data against concurrent loop detector stations over several months, Transportation Research Part C: Emerging Technologies, Volume 49, 2014, Pages 59-72, ISSN 0968-090X, https://doi.org/10.1016/j.trc.2014.10.002.

[Seo2019] Toru Seo, Yutaka Kawasaki, Takahiko Kusakabe, Yasuo Asakura, Fundamental diagram estimation by using trajectories of probe vehicles, Transportation Research Part B: Methodological, Volume 122, 2019, Pages 40-56, ISSN 0191-2615, https://doi.org/10.1016/j.trb.2019.02.005.

Welcome with your application!
Does this sounds interesting and you would like to know more, please contact Ian Marsh, Ph.D, +46707721536. Last day of application is the 27th of May. 

About the position

City

Kista

Contract type

Temporary position 3-6 months

Job type

Student - Master Thesis/Internship

Contact person

Ian Marsh
+46707721536

Reference number

2020/129

Last application date

2020-05-27

Submit your application