Friday, December 26, 2014

PhD Position in Complex Networks in Transportation

We're seeking applicants for a funded PhD position (3-year) in the Department of Civil Engineering, Institute of Transport Studies at Monash University, Melbourne, Australia. The general topic of this PhD work is on the application of complex networks theory to understand and model urban transportation systems.

Qualifications of the Applicant:
We are looking for a highly motivated and qualified PhD student with interest in the fields of complex networks and transportation modeling. The applicant should hold a Master's degree in applied mathematics or engineering. He/she should have a solid background in mathematics and programming in Matlab. Applicants with at least one or two peer-reviewed journal publications are preferred.

Applicants should send their CV via email to meead [dot] saberi [at] monash [dot] edu.
To learn more about the city science research group at Monash University, please visit:

Position is open until filled.

Wednesday, December 17, 2014

Prof. Serge Hoogendoorn and Dr. Winnie Daamen visit ITS Monash

Colleagues from Delft University of Technology, Prof. Serge Hoogendoorn, Dr. Winnie Daamen, and Dr. Sascha Hoogendoorn-Lanser (from The Netherlands Institute for Transport Policy) meet with ITS students and faculty at Monash.

Friday, November 21, 2014

Professor Elise Miller-Hooks visit ITS Monash

Professor Elise Miller-Hooks from University of Maryland meet with ITS students and faculty at Monash University.

Wednesday, November 19, 2014

"Peak Car" in Australia

Following a discussion on young people driving trends in Australia posted on the ITS LinkedIn page by Alexa Delbosc, one of my colleagues, here I am posting a few graphs showing an apparent "Peak Car" in Australia.

The total VKT (in million) between 1965 and 2012 in Australia is shown first. What I see in this graph is an increasing trend. However, if you look more closely you may notice that the increase rate is actually going down a little bit (the slope becomes smaller).
The slowing down rate is more obvious when I plot the change in total VKT from year n-1 to year n as shown below. The change in VKT has been fluctuating year to year. However, I see a general decreasing trend there.
And now if we divide the total VKT by total population to get the VKT per capita, we can see a peak in the graph happened in 2004. This is 3-4 years before the Global Financial Crisis (GFC)!
This is actually very interesting because the decline in VKT per capita in the US started in 2006, about 1-2 years before the GFC (reference). It was also in 2006 when the American housing market had a burst with house values peaking (reference).

Did a same thing happen in Australia? Is there really a relationship between housing prices, income, and people's driving trends? I'll get back to this question with some data from Australia in a bit.

Here is the change in VKT per capita from year n-1 to year n. I see a similar decreasing trend with many fluctuations.
Going back to the question on the relationship between housing boom and driving trends, here is also some data from Australia showing a "local" (not global) housing boom in 2003-2004 about the same time as the apparent "Peak Car" happened. The graph below shows the ratio of median residential house price over the annual wage & salary in Australia. Data is from ABS and REIV.

Does this mean people started shifting their behavior towards less consuming (fuel/driving) because they were too in debt on their housing costs? I don't have a certain answer yet. After a local boom in housing prices in 2004, there has been a relatively stable (on average) period until more recently.
The other question which still remains open is "will this trend continue?" To answer this question, we need to understand why we see what we see. If we figure out the driving factors that push VKT up or down, then we may be able to forecast the future. Forecasting is a difficult task.

UPDATE: Following you'll see a few more graphs for fuel price, fuel budget, unemployment rate, and young adult (18-34 years old) proportion.

What is fuel budget?
Fuel budget is defined as the budget that a person spends on fuel in a year.

How did we estimate fuel budget?
Fuel budget = fuel price [$/liter] x average fuel consumption [11.5 liter/100 km] x VKT per capita [km/person/year]

Tuesday, November 18, 2014

The Cost of Traffic Jam

A recent article on The Economist on cost of congestion, based on a report by INRIX:

"The Centre for Economics and Business Research, a London-based consultancy, and INRIX, a traffic-data firm, have estimated the impact of such delays on the British, French, German and American economies. To do so they measured three costs: how sitting in traffic reduces productivity of the labour force; how inflated transport costs push up the prices of goods; and the carbon-equivalent cost of the fumes that exhausts splutter out.
In America the average cost of congestion to a car-owning household is estimated to be $1,700 a year; in France it is $2,500. But traffic is so bad in Los Angeles that each resident loses around $6,000 a year twiddling their thumbs in traffic—at a total cost of $23 billion, the costs are estimated to exceed that of the whole of Britain. But these costs do not take account of the price of carbon-dioxide emissions. In total, over 15,000 kilotons of needless CO₂ fumes were expelled last year—which would cost an additional $350m to offset at current market prices. In choked-up Los Angeles $50m alone would have to be set-aside."

Thursday, October 9, 2014

The Guardian publishes Melbourne Bike Crash Map

The Guardian just published our Melbourne Bike Crash Map:
"Researchers from the Institute of Transport Studies at Monash University have mapped five years of cycling accidents in Melbourne. The red circles are crashes, the density of the blue layer shows the number of cyclists based on information from the cycling app Strava. You can read more about the map and see more graphics based on the same figures at the Monash ITS site."

Herald Sun features Melbourne Bike Crash Map

A Herald Sun article on "Cyclists want chunk of Chapel St to be a bike-only zone" features our Melbourne Bike Crash Map.
"Monash University’s civil engineering department has developed an online map showing bicycle crashes in Melbourne, using VicRoads data from 2008-13. It showed 6219 cycle crashes in that time. Beach Rd tops the list, followed by Nepean Highway, St Kilda Rd and Sydney Rd. Chapel St was sixth on the list, with 91 accidents."
Link to the article:

Friday, September 5, 2014

Melbourne Housing Density Map

Melbourne has more than 1.5 million private dwellings. About 9% of them are unoccupied, according to Census 2011. Of occupied private dwellings in Melbourne, 73% are separate houses, 15% are flats/units/apartments, and 12% are semi-detached/townhouses. With the rapidly growing population, there is obviously a need for more housing. There are two general housing development trends in Melbourne: 1) building more high rise apartments in the inner city and 2) more housing developments on the urban fringe. Each has its own advocates and critics.

This project <> uses household dwelling data at the Mesh Block level from Census 2011 to visualize the spatial distribution of housing density in Melbourne.

Tuesday, August 26, 2014

How socio-demographic changes may impact travel demand?

A recent NCHRP report released on 19/08/2014 discusses the effects of socio-demographic changes on future travel demand.

Wednesday, August 6, 2014

Top 20 Streets in Melbourne with Highest Number of Bicycle Crashes (2008-2013)

Following our latest visualization on "Melbourne Bike Crash Map", I have put together a list of the top 20 streets in Melbourne with highest number of bike crashes between 2008 and 2013.

Note: Data shown below is the number of bicycle crashes only. It is NOT normalized by the length of the road or the bicycle volume.

Friday, August 1, 2014

Melbourne Bike Crash Map

Link to Melbourne Bike Crash Map:
"Cycling is growing in Australia. The number of bike trips to work in the Greater Melbourne area has increased from 12,124 in 1991 to 25,572 in 2011. Despite the growing interest in cycling, bike safety has remained a concern. 
In this project, we visualize 5 years' worth of bicycle crashes from 1/7/2008 to 1/7/2013. We used the Strava Global Heatmap as the background layer. Strava is a cycling app which allows users to track all their rides. The Global Strava Heatmap includes more than 200 billion data points from users showing where people (or more precisely Strava app users as an imperfect sample of population) ride. The number of bike crashes alone does not give a complete picture of the safety characteristics of a location. Crash rate is actually a better measure determining the relative safety considering the bicycle traffic volume. Since bicycle traffic volume data is scarcely available, we have used the Strava Heatmap to fill the gap."

Wednesday, July 30, 2014

Melbourne Age Dot Map

We just released the "Melbourne Age Dot Map", visualizing the spatial distribution of population by age in Melbourne. A pattern that stands out most is the concentration of Millennials (20-35 years old) in the CBD and inner city areas. Also, a closer look at the newly developed fringe suburbs reveal that those areas are mostly inhabited by population under 65 years old.

Saturday, July 19, 2014

Transportation and Politics in Australia

I'm not much into Australian politics but it's interesting to see how transportation is playing a key role in local elections. "Only Labor will fix Chandler Highway Bridge." versus "Only Liberal will build the Melbourne airport rail link." I actually think both the new Chandler Highway Bridge and the Melbourne airport rail link need to be built, no matter who is in power. The new Chandler Highway Bridge will have more local impact while the new Melbourne airport rail link will have more city-wide impact. A through cost benefit analysis needs to be done to prioritize such large investments.

Friday, July 18, 2014

PhD position in the area of Machine Learning, Optimization, and Visualiation

I have another PhD position just recently opened in my research group in the area of Machine Learning, Optimization, and Visualization. If you have electrical or computer engineering (software engineering) background and interested in this opportunity, please email me at <meead [dot] saberi [at] monash [dot] edu>. Students with some experience with JavaScript, HTML, Python, and optimization are preferred. The PhD position will be funded by NICTA <>.

Application Deadline: Open until filled (August 2014 - January 2015)

Friday, July 11, 2014

Racial Segregation: Chicago, Washington D.C., and Melbourne

Comparison of racial/ethnic diversity in Chicago, Washington D.C., and Melbourne

In all three maps, blue color refers to white. Other colors such as green, orange, red, yellow, etc represent non-white (e.g. Asian, Hispanic, Black, etc.). Chicago actually consists of largely segregated pockets of black, Hispanic, and white people. Washington D.C. is even worse where half of the city is white, the other half is black. However, Melbourne seems pretty mixed. Note that white is the dominant race in Melbourne (more than 70%) and non-white population in Chicago and D.C. are much larger than in Melbourne. As the non-white population in Melbourne increases, I can't tell whether Melbourne will remain mixed in the coming decades or will take the same path as some American cities had taken.

Media Appearance: Melbourne Language Diversity in "The Age"

The Age article today on diversity of spoken languages in Melbourne refers to our visualization of ethnic distribution. I wish they included a link to it too.

Wednesday, July 9, 2014

Visualization of Air Travel in Australia

Our third visualization project on "Air Travel in Australia" is online now. In 2013, there were 87 million passengers (domestic and international) carried in the Australian airspace while the population of Australia is only 23 million!

Link to the interactive visualization: 

Friday, July 4, 2014

Open positions in Transportation Engineering at Monash University

We have two new open positions in Transportation at Monash University at the lecturer/senior lecturer level which is equivalent to assistant/associate professor level in the US system. Preferred areas of research include (but not limited to) land use and transport modeling, activity based modeling, and freight modeling. Application deadline is July 20, 2014.

Here are links to the job advertisements: (Lecturer) and (Senior Lecturer)

 Great opportunity to experience life and work in Australia! You'll enjoy it.

Monday, June 30, 2014

Melbourne Population & Dwelling Density Distribution (2011)

According to Census 2011, Melbourne population (as of 2011) is 3,999,982. There are also 1,572,171 private dwellings in Melbourne metro area. Melbourne covers an area of 9990 square km. Therefore, average population density in Melbourne is 400 person/sq-km and average dwelling density is 157 dwellings/sq-km.

However, the spatial distribution of population and dwelling density is not uniform. Following figures show how population and dwelling density vary as we move away from city center. I drew multiple 10 km rings in ArcGIS overlaid on top the mesh block and population data. Population and dwellings in each ring are counted and divided by the sum of mesh block areas to get the densities.

Dwelling densities vary from more 1800 dwellings/sq-km to near zero. Similarly, population densities vary from more than 4000 persons/sq-km to near zero. Obviously, the average densities decrease as we go farther out from the city center. However, this does not necessarily mean a neighborhood in the middle of the city is always denser than a neighborhood in an outer suburb. One reason that outer suburbs have much average lower density is the existence of vast amount of undeveloped (or non-residential) lands between blocks/neighborhoods.


* UPDATE (03/07/2014)

I have also plotted "ring maps" of dwelling and population density in Melbourne. Starting from a point in CBD, several rings with radius incrementally increasing by 5 km are overlaid on population and dwelling data from Census. The following ring maps may provide a better picture of the spatial distribution of density in Melbourne.

Friday, June 27, 2014

Melbourne Ethnicity Dot Map

We just released our second visualization project "Melbourne Ethnicity Dot Map" trying to show the extent in which Melbourne is ethnically mixed/segregated. The dominant race/ethnicity is obviously white from West Europe, Australia, and New Zealand. Except a few suburbs with non-white (non-West Europe) majority, the rest of the city looks pretty well mixed if you zoom in to the max level. The point is, Melbourne is not as segregated as Chicago or D.C. at all.

Tuesday, June 24, 2014

Changing Melbourne: Population Dot Map of Melbourne (2006-2011)

We just released our first visualization project. The project includes a few informative graphs and a dot map of population (change) in Melbourne. Every dot in the map is a person. We've generated more than 4 million dots to replicate population distribution using Australian census data.

Changing Melbourne: Population Growth

Thursday, May 15, 2014

Melbourne needs Uber!

While the "sharing economy" is growing worldwide, the controversy around ‪#‎Uber‬ has gone wild in ‪#‎Melbourne‬! The new technology-based competition is shaking up the more traditional markets which haven't been changed for a while. I think cities should embrace opportunities like Uber and perhaps cooperatively work with them to make sure all users will enjoy a safe service. Lack of competition in taxi services in many cities has created an unpleasant environment (in terms of cost, quality of service, availability, etc) for users. It's time to re-think about some of the traditional markets and regulations around them.

See this article on Sydney Morning Herald on the recent crackdown on Uber drivers:

Tuesday, March 11, 2014

PhD positions available in City Science Group at Monash University

I have launched my research group webpage today: City Science Group. The group has different research themes including urban transportation modelling & analytics, understanding human mobility patterns in cities, urban big data, multi-modal urban transportation planning and operations, bicycle and pedestrian innovations, and urban freight and logistics.

There are currently a few PhD positions available at my group. Qualifications include BS/MS in Science, Engineering, or Mathematics. Applicants with a Master's degree with one or more of the following backgrounds are preferred: transportation network/traffic modelling, operations research (optimization), complex networks, and online data visualization.

Thursday, March 6, 2014

Zipf's Law for Australian Cities (Significant Urban Areas)

Following a post on the verification of Zipf's law for cities in Iran, here I am testing whether the Zipf's law also holds true in Australia where the growth of cities has not been completely natural with many historical and political changes, mainly immigration.

This is a chart showing the Australian significant urban areas ranked by population. Melbourne and Sydney's population are very close which contradicts the Zipf's law that the second city in the ranking list should have half population of the first city. This might be due the man-forced changes happened throughout the years in Australia and the immigration policies. 

However, if we look at the rest of the cities, they closely follow the Zipf's law. The following figure shows the log population versus log ranking of cities in Australia. Excluding Sydney from the analysis, Melbourne, Brisbane, Perth, and other cities line up nicely for a power law to fit. Of course, further investigation is needed to better understand the population growth patterns in Australia.

Saturday, February 15, 2014

Melbourne at First Sight

It's been two weeks that we have arrived in Melbourne. Melbourne isn't very different from large cities in the U.S. In fact, to me it's a combination of Portland, San Francisco, Chicago, and Miami. It's a sprawled city with great public transportation and biking infrastructure with beautiful beaches. The public transport in Melbourne consists of three main types/modes: trains, trams, and buses. The tram network is huge. The train network is radially shaped, just like Chicago with a city loop. It's very hard to get from a suburb to another suburb with public transit though. There seems to be numerous opportunities for research. Just like New York in the U.S., Melbourne is a live transportation research laboratory in Australia.

Tuesday, January 28, 2014

Improved Public Transit in Mashhad, Iran

One of the nicest improvements I've noticed in Mashhad is the new bus stops. They all now look the same and include a metal-glass shelter, bench, and a big map of the route(s) (a schematic view of the stops and an actual city map). The bus loading area is marked with yellow zigzag stripes. The bus fare is paid electronically in the bus with a contactless smart card. No timetable or bus frequency information is yet shown in the stop though.

Thursday, January 16, 2014

Presenting NFD/MFD at Mashhad Traffic and Transportation Organization, Iran

Presenting some of our recent findings on network level traffic flow dynamics (NFD/MFD) at Mashhad Traffic and Transportation Organization, Iran. Thanks to Ehsan Jamshidi for organizing the seminar and the pictures. Also, thanks to Hani Mahmassani and Ali Zockaie for the research work and collaboration.