EyeforTravel Amsterdam 2018

November 2018, Amsterdam

Own your Data. Master Digital. Dominate Bookings

Learn from Travel Industry Decision Makers how to Thrive in a Platform Dominated World

Data science no doddle, says Trainline

Investment in data is helping Trainline to build innovative solutions that deliver customer value in the fast-growing rail sector. Pamela Whitby reports

In the past year alone, the data team at Trainline has grown from ten to around 50 people.

“All of that data investment,” says Fergus Weldon, Trainline’s director of data science, and EyeforTravel Amsterdam keynoter, “goes into building innovations to make journeys simpler, smarter and smoother for travellers”.

Across the rail industry, investment is accelerating, and not surprisingly. According to US-based Grand View Research, the rail market is growing at around six per cent a year, and is forecast to reach $830 billion by 2025. Asia, particularly India and China, are pouring money into lines, and everywhere, rail is being touted as a less stressed and more environmentally friendly way to travel.

The rail market is growing at around six per cent a year, and is forecast to reach $830 billion by 2025

On the tech front, much work is currently being done to accelerate the capabilities of data science in travel. Greater data sharing between transport providers and tech companies is taking place, and is even being encouraged by policy makers. In this, rail has not been missed (as the EyeforTravel Amsterdam speaker line up indicates). In the UK, for example, the government is directly working on this through its Joint Rail Data Action plan with the Rail Delivery Group, which is designed to enable more intelligent use, and greater sharing, of data across the industry.

Says Weldon: “Our own advances in data science have led to innovations such as the UK’s first price prediction tool for rail tickets, which has already saved customers £9 million.” 

Crowdsourced seat-finder feature BusyBot is another result of Trainline’s data driven efforts, as is its recently launched AI-powered voice disruption alerts. In what it claims is a world first, the system reads train operators’ tweets on disruption and automatically shares voice updates to relevant customers if their journey has been affected.

Obvious and not so obvious skills  

Put simply, data science is the application of scientific principles to build innovative applications, and that is exactly what Trainline, a company that sells rail and coach tickets for 183 companies worldwide, is doing.  

However, data science is no doddle, and finding talent that ticks all boxes isn’t either. Of course, there are number of essential skills required for the discipline, some obvious, and some not so obvious. 

Problem solving is naturally an essential skill that any scientist needs. But data scientists also need to be curious, passionate and determined. The work is not for the fainthearted because while 80% of a project may be straightforward, the final 20% is often full of curve balls and extremely challenging,” says Weldon.

It is in this last mile that individuals need real drive to push through and deliver beneficial impact for customers, and that’s not always easy to find.

Although, perhaps, not automatically associated with data scientists, another important skill is storytelling.

“If you can’t tell the story of the data you’re working with and the innovation it is driving - both externally and internally to key stakeholders – you’ll quickly find the business stops paying attention to the importance of data science,” warns Weldon.

To stay on top of industry trends, and to find the best skills, Trainline’s growing team of data scientists get involved in external meet-ups throughout the year and speak at industry events like EyeforTravel Amsterdam. “We’re embedded in the international tech community. That way we get to meet new talent from across the world and continually tell them about the work of our team,” he says.

Getting the foundations right

There is a view that, as Humayun Sheikh, a former lead investor in Google DeepMind and now co-founder and CEO Fetch AI, puts it: “The only companies [aka the tech giants] able to run effective AI algorithms and gain insights from their data are those with massive data stores and the opportunity to invest large sums in the right people and tools to analyse them.”

It’s short-sighted to think the only companies that can effectively innovate [with data] are in Silicon Valley

Weldon acknowledges that the tech giants, aka Google, Facebook and Amazon, are able to do amazing things with big data but argues: “It’s short-sighted to think the only companies that can effectively innovate [with data] are in Silicon Valley”.

He continues: “The biggest tech companies may have the deepest reserves of data, but wherever a company has quality, relevant data, innovation can take place. Even businesses that don’t have the scale of Google and others will be required to pull unique insights from their own unique data sets to stay ahead of the curve.”

Aligning aspirations, ensuring quality  

So what is the trick to becoming more effective with data? Weldon believes firms need to ensure their attitude towards data is always aligned with their aspirations. “AI is at the top of a pyramid of capability, it doesn’t work without the foundations,” he says.

Anyway, Trainline isn’t short of ‘big data’ in the truest sense of the phrase. The firm helps customers take over 127,000 journeys every day, sells over 175 tickets every minute to people in 173 countries around the world.

“When it comes to travel and journey planning we are in a fantastic position to build machine-learning models that use this data effectively,” says Weldon, who argues that in data science data is the fuel, and if it’s of low quality then the work will be too. The worst thing that can happen, he says, is investing time and effort into building something from data that turns out to be of poor quality.

To avoid this, Weldon, who will share more in Amsterdam, has this advice.

  • Minimise the number of assumptions being made about the quality of the data you work with
  • Put the right checks in place in order to understand the full context before starting out
  • Understand third parties 

“When you work with a wide variety of third-party data (such as that from train or coach operators), it can be altered without you always being aware,” says Weldon. For this, Trainline build anomaly detectors – essentially, algorithms that understand the make-up of the data and monitor for any changes over time. So, if things look unusual, relevant members of the team can be alerted, and investigate.

In just short of two months, more than 300 travel executives and innovators will gather for EyeforTravel Amsterdam, where Fergus Weldon, Director of Data Science, Trainline, will be speaking

Related Reads

comments powered by Disqus