Solving Traffic Congestion: What We Built at Grab’s 24-hour Hackathon

I remember I was wandering nervously around the warm, welcoming cafe area at the ADAX (ASEAN Data Analytics eXchange) office right after registration — trying to search for some familiar faces.

The first stranger I talked to was Choong, a software engineer at MapIt. We greeted each other and exchanged name cards. It turned out, he did not have a team yet.

Coming to a hackathon without a team, I could feel the anxiety. Most of the people were sitting in clusters at the dining table. Some even brought their computer monitors in luggage bags. Truth is, I did not know what would be the outcome, in part which makes it exciting. Right after the briefing, I approached the organizer and asked if there is a team formation session. I was told there is none but I was referred to Sim, a data analyst working in a local bank, who was looking to join a team as well. I met Thomas during lunch, he is a web developer — and together finally, we formed a team to hack for the next 24 hours.

📸 : Event Photographer and Choong

Framing the Problem: How Might We Help Manage Malaysia’s Traffic Better?

We were given the OpenTraffic data set to work on, with the interactive Carto map and the Mapillary street level photos as the potential tools for solving the problems. We picked problem statement 1, which is to use the dataset provided to help the government plan more effectively and make informed, evidenced-based decisions on traffic interventions.

Traffic congestion really is a problem that everyone hates. The noise, the pollution, the frustration and the wasted hours sitting in the car rather than getting home and spending quality time with family. According to a World Bank study in 2014, traffic congestion causes RM20bil in economic losses in the Klang Valley alone — that’s RM54mil a day. Besides, traffic jam also causes Malaysian to waste one million hours per day.

The Opportunities: The Road is Always Greener On The Other Side?

The opportunities that we saw, really lies between the empty roads in the opposite direction that we can utilize during the traffic jam at peak hour.

It turns out, we already have a solution. Above is a photo captured on Google Maps on the road segments at Jalan Cheras heading towards the city center. Every once in a while, you will see the road blocks instead of the road divider when you are driving, and Malaysian authorities have implemented the rerouting of traffic through opposite direction roads during peak hour. #MalaysiaBoleh

Below is a visualization I made, to demonstrate how it looks like:

During peak hour in the morning and evening, there are a few main, long stretches of roads connecting the KL city center and the residential areas, up to 10 km or more. But currently, the authorities have no accurate idea on when is the optimized timing to implement the road blocks (is it 4–7pm, or 3–8pm?), which optimized road segments or routes to implement, and what are the new routes to expand on in order to ease the traffic congestion.

The Solution

By observing if there is a peak in the distribution of the average deviation, we can analyze the time slots and which segment of roads are more susceptible to jam. Thanks Sim for coming up with the formula:

After cleaning up the datasets, we imported them into Carto, overlayed and analysed as below:

By using the map visualization, authorities can now analyse the road situation and take actions based on evidence-based data sets by setting up roadblocks based on optimized timing, optimized road segments and new routes to expand.

Future Improvements on Dataset

**_Missing gaps in between road segments_**Based on the data provided, massage the data to identify the connectors between the main road segments with missing informations (average speed)

**Outliers in the data**Remove the outliers due to the GPS ‘jumping’ problems before the aggregation of data to obtain average speed for the road segment

Key Takeaway Tips for Beating Traffic Congestion

Until innovative solutions that address the systemic issues are implemented, here is what we can do as drivers on the road: maintain a constant speed that is perceived to be safe and try not to change lanes unnecessarily, in order to achieve an optimal flow of traffic. Traffic engineer Brian Wolshon explained about traffic shockwave in this simulation:

We did not take home any prize this time, but it was a great learning experience for me using Carto and data analytics nevertheless. Congrats to team KingsoftheNorth, Ditto, S.I., Rendezvous, and FTT who won the top five — amazing hacks and well-deserved! Special thanks to Grab, MDec, Carto, Mapillary and OpenTraffic for organizing the hackathon!

Thanks for making to the end of this journal! Are you enjoying reading it? What are other innovative ideas that you might have thought of to solve traffic congestion?

Let me know by commenting below ⇣ or sending me a tweet on Twitter 🐦 and I’d be excited to join the conversation. Thanks for hitting the 💚 if you enjoyed this article!

Chip Dong Lim is a designer currently working on CarinMED — a simple, affordable electronic medical record that helps doctors efficiently and effectively manage the medical histories of homeless and urban poor patients. View more of his past design work at madebychip.com.