Using Big Data and Data Science to Solve Traffic Congestion Woes in Developing Countries - TenPoint7
18159
post-template-default,single,single-post,postid-18159,single-format-standard,bridge-core-3.0.1,qode-page-transition-enabled,ajax_fade,page_not_loaded,,qode-title-hidden,qode-theme-ver-30.4,qode-theme-bridge,wpb-js-composer js-comp-ver-6.7.0,vc_responsive

Using Big Data and Data Science to Solve Traffic Congestion Woes in Developing Countries

It’s common knowledge that traffic management in developing countries is a tough problem to solve for multiple reasons, some of which are financial and political. Fortunately, solving this challenge purely from a technical or scientific perspective is much easier than the other dimensions.

In this blog series, I will take this opportunity to share some of my insights gathered while developing an Intelligent Transportation System (ITS) for the city of Ho Chi Minh City (HCMC). Over the course of the next few blogs, I will share some of the statistical techniques that were used and also how “big data” was utilized to solve this challenging problem, many of which are also applicable in solving real world business problems.

As many of you might know, HCMC is one of the biggest and most traffic-active cities in SE Asia with very complex traffic systems and patterns including a high density of motorcycles. Moreover, many roads in the city center are non-lane-based making them hot spots of congestion, which only gets worse during peak rush hours.

Consequently, the complexity of collecting traffic data becomes exponentially tough. An Intelligent Transportation System (ITS) therefore becomes a potential solution for dealing with such complexities. As a matter of fact, ITSs have been built and deployed in developed countries under different formations. However, they are rarely used in most developing countries primarily due to the cost of creating, implementing and maintaining such systems.

At the John von Neumann Institute, we proposed the design of a constructive ITS suitable for developing countries, such as Vietnam, that makes use of GPS data collected from multiple transportation modes. This was an undertaking sponsored by the city’s Ministry of Transportation.

There were 3 main business/operational goals of the ITS project:

  • Ease of Deployment

The main sources of data for the ITS system is GPS data sourced from available devices such as on-board GPS devices, traffic cameras, traffic sensors, etc. as well as end-user’s smartphone GPS data. As such, no system installation for data collection is required in this model, which is common in other ITSs. This made our system more deployment friendly. However, this also meant that we needed to establish a criterion to standardize the collected GPS data and then convert it to a unified class for downstream analysis (more on this in a later blog post).

  • Cost Efficient System

Like any large-scale system, the overall solution needed to be cost efficient for development and ongoing maintenance. In this case, we established that the cost of developing, implementing and maintaining our system by the operator (i.e. Ministry of Transportation) was much lower than existing ITS models. We also determined that the solution needed to be user-friendly and easy-to-use for end-users. Fortunately, the procurement cost of GPS devices for motorcycles in countries like Vietnam is low and adoption is mostly widespread, therefore such devices are a very accessible source of data. Furthermore, prices of smart devices (such as smart phone, tablet, etc.) are becoming more affordable and popular in HCMC today. Lastly, costs for developing mobile applications (for collecting GPS data) are becoming increasingly cheaper.

  • Effectiveness

It has been experimentally suggested that 2−3% penetration of GPS devices in total drivers population is enough for estimating velocity of traffic flow quite accurately and efficiently on certain road segments. In developing countries such as Vietnam, there are a significantly large number of small roads that are only suitable for motorcycles. Therefore it was decided that the use of GPS motorcycle data and mobile application GPS data could sufficiently enable our ITS to cover most urban road networks in HCMC.

Aside from the business goals, on a technical level, our ITS solution aimed to achieve two broad objectives:

  1. To collect and efficiently process traffic data from various types of sources, as mentioned earlier, such as cameras, sensors, GPS on cars, buses, taxis, motorcycles, etc. or via individual users through our own mobile app.
  2. Adapting relevant & existing algorithms in transportation research to regulate traffic flow in congested cities and informing end-users via a control center.

In the upcoming blog posts, I will detail some of the technical challenges that we faced – starting with data analysis and pre-processing.

I look forward to continue sharing some of my insights and also look forward to hearing your comments.

An Mai, PhD (Data Scientist & Researcher, An@TenPoint7.com)



TenPoint7