Data, Artificial Intelligence (AI) and Machine Learning (ML) are the three hottest buzzwords at present. Every organization is attempting to search and develop their own information systems to be smarter and more efficient by hiring data scientists to gather massive amounts of data to analyze. This is in order to enhance their services and develop our society.

Techsauce talks to Grab’s head of Data Science (Machine Learning) Jagannadan Varadarajan who is based at Grab’s headquarters in Singapore. He is the director of Grab-NUS Lab, a collaboration between Grab and the National University of Singapore (NUS) to develop Singapore’s transportation system. This interview gathers various perspectives and offers clear examples on how information can be used to benefit business and society.

How has Grab used the information it has at present to create solutions?  Do you have plans for the future?

Traffic congestion is one of Thailand’s biggest problems. In fact, Bangkok was ranked as the most congested city in Asia in 2017. Fundamentally, with Grab available we hope to give more people less reason to own a car. Furthermore, every day, Grab’s fleet of drivers are crisscrossing roads, big and small, in Southeast Asia. We drive every road that matters multiple times a day. The Grab driver application produces GPS data, allowing us to collect a massive trove of information. The data is so rich that when the GPS points are put together, they look like a map of the city.

We are using this data to build better Machine Learning and Artificial Intelligence models to predict traffic congestion, traffic incidents and travel times. We also use historical and real-time demand and supply changes to predict these signals upfront in time. This, in turn, helps us do things like prompt our drivers to move to areas of high demand so that they get their jobs and people get their rides.

Beyond that, we are also working with government agencies and other organizations to use our traffic data to solve traffic and road safety problems across Southeast Asia. For example, we contribute aggregated, anonymized driver GPS data to the OpenTraffic platform. The data from that platform could be used by traffic authorities and city planners to ease congestion patterns on roads by improving traffic signal times or route planning. City planners could also make better decisions on where to build road infrastructure.

In partnership with the National University of Singapore, we have set up the Grab-NUS AI Lab to create a robust AI platform for large-scale machine learning and visual analytics that can develop novel applications from Grab’s massive data set. For example, assisting transport authorities in monitoring and optimizing traffic flow.

We are also looking at how shared ride services can be sensibly implemented in Thailand to encourage more carpooling and fewer cars on the road. For example, our data shows that travel time for Pratunam Market to Don Muang Airport can be drastically improved. If this route would be better served by more shared transport solutions, such as buses, trains, ride-sharing, we could bring travel time during peak hour down by 25% all the way from 45 to 34 minutes.

Have you implemented any ML and AI in real projects?

Grab has a lot of data. In fact, we generate 20 terabytes of data daily. But having the data is only one half of the equation. Equally important is having the tools and capabilities to transform this data into intelligent features that can be used to continually improve the customer experience and even transform smart cities of the future.

This is where data science comes in, and it’s why Grab has invested heavily in AI, machine learning and operations research.

• In the area of payments, we use the driver behavioral and earnings data to develop a credit scoring model that assess the creditworthiness of drivers. This helps us to extend a line of credit to drivers (who do not get loans through conventional banking facilities) to address their daily needs such as servicing their car, buying fuel or paying up their utility bills.
• In food delivery, we use AI to understand customer preferences so that we can recommend restaurants and food items that are localized, trending and tailored to their palate. We also use simulation and optimization to improve matching efficiency every single day, so that delivery-partners can minimize travel time and consumers can receive their food quickly.
• In transport, we use AI to accurately estimate travel times and improve our maps, using vast amounts of GPS trails to infer and correct our map geometry. One new feature we recently rolled-out enables drivers to complete as many jobs as possible between their present location and their homes. It matches drivers with passenger destinations that are along the way so that drivers do not end up traveling long distances home with an empty car and a missed revenue opportunity.

Can you give an example of A/B testing that has been successful?

At Grab, we continuously strive to improve the user experience of our app for both our passengers and driver partners. To do that, we’re constantly experimenting, and in fact, many of the improvements we roll out to the Grab app are a direct result of successful experiments. A/B tests are an important feature of experimentation. We use A/B tests to determine which of two or more variations, usually minor improvements, produces the best results.

We use A/B testing in many areas, including search and recommendation algorithms, pricing & fees, site architecture, outbound marketing campaigns, transactional messaging, and product rollouts.

How can information gathered by Grab be used to benefit the government?

The data from the various services we offer allows us to gain extremely precise insights into:

• People’s commute patterns: This helps to understand how people commute at different times of the day, which can be used to better plan the city’s infrastructure, public transport, and formulate better traffic policies.
• Traffic scenario in every road segment of the city: this can help governments to predict travel times and understand the impact of traffic so that they can implement methods to ease traffic congestion during peak hours.
• Demand and supply: This helps to understand where people congregate at different times of the day and their preferred modes of transportation. This information could be used as inputs to plan and organize different neighborhoods.
• People’s eating patterns: As we move into more services such as food, we can also use information about the popularity of restaurants, cuisine types and food types to understand people’s eating patterns, which can be used to promote health and hygiene, control quality and minimize wastage of food.

How has Grab worked with the government to manage data?

We actively work with governments and regulators across Southeast Asia to use the multi-petabytes of data collected on our platform to develop and optimize services for Southeast Asian cities. Apart from providing our driver location data to the OpenTraffic platform which is a collaboration with the World Bank to provide Southeast Asian governments, currently in Myanmar, Indonesia and the Philippines, access to real-time traffic information, we are also exploring how to use Grab’s data to help governments directly with transport planning, complement unmet demand in transport and map out how car growth affects cities.

What is Grab’s mission in creating a smart city?

Grab is committed to driving Thailand towards a smart city future and has been collaborating with partners from various sectors, the government included, to achieve this goal. A prominent example is the MotoGP event in October 2018 in which we collaborated with the Buriram Provincial Government Office to pilot the Smart Mobility project to drive the province towards a Smart City future. The move aligns with the “Buriram Model” vision which aims to boost the province’s economic and tourism growth while sustainably improving the incomes and quality of life of local residents. In the near future, we are looking to collaborate with other provinces, especially secondary tourist cities, in a similar manner.

Last year, Grab also launched the Safer Everyday Tech Roadmap in Thailand with the aim of making Thai cities smarter and safer. The key element of the roadmap is the enhancement of road safety standards to eliminate preventable incidents by monitoring drivers’ fatigue from in-house telematics reports and encouraging a change in users’ safety habits which could lead to long-term road behavior change.