The 'beast from the east' may have reduced many of the UK's roads and railways to a standstill, but 200 leaders in data and analytics nonetheless made it through the snow to the Oxo Tower in London for the launch of the DataIQ 100 list of the most influential people in data last week.
Many may not have made it if not for the work of a data science team led by the woman who ranked seventh on the list, Lauren Sager Weinstein.
The chief data officer at Transport for London (TfL) and her team analyse a vast volume of data from 1.37 billion annual passengers to plan services across a 402km network.
"What we want to do is we want to make sure that we're giving information back to customers," Sager Weinstein tells Computerworld UK as the snow outside covers the south bank of the river Thames.
"That's hugely important and I think that's transformed the way that people travel around London."
Giving information back to customers
When Sager Weinstein joined TfL 15 years ago, the travel decisions made by passengers were largely guided by guesswork. The proliferation of data and mobile internet means they can now have real-time transport information beamed to a device in the palm of their hands.
The TfL of today amasses huge datasets across Oyster and contactless payment cards, ticket sales, vehicle sensors, surveys and social media and information from external sources such as health and housing organisations.
TfL searches for patterns in this data to understand which people are using the network, when they're travelling, where they're going and how they're getting there. They can then operationalise the information into services that support users and the network through smooth service periods, and interruptions such as the recent snowstorms.
On occasions, these incidents cause an outage that suspends or terminates a journey. If it's delayed by 15 minutes or more for a reason within TfL's control, customers can apply online for a service delay refund. TfL then analyses data on the route to assess the validity of the claim.
It can also proactively identify the affected customers.
"They don't need to claim for a refund on our website, we just pay the refund out them and it goes back to their Oyster or their contactless payment card," says Sager Weinstein. "That's one area where we give a direct service back to our customers."
Sager Weinstein admits that not every problem on the network can be solved by analytics though. Maintaining a service through the snow has only been possible thanks to the efforts of a wide range of staff across TfL.
"I do think it's important to say that not every problem is an analytics problem," she says. "Through the snow, we've been running our services well. We've been prepared, but that's not just analytics, it's just good operational preparedness."
Analytics nonetheless looks likely to play a growing role at TfL. Trials are underway that add predictive maintenance to trains and explore the potential of AI, but Sager Weinstein wants to ground emerging technology in contemporary problems.
She previously worked in public policy in Los Angeles, and before that studied Public and International Affairs at Princeton University, giving her a strong understanding of the needs of citizens.
"We need to then also give a little bit of space for our data scientists to play with the more cutting edge, and it's a balance that I have to reach," she says. "You want your data scientist to be creative but you also want help with some of the most pressing business problems."
Service planning with analytics
TfL's data team also analyses where customers are going on the network to plan the services by comparing Oyster usage with bus journeys.
"We have analytics that can make an inference on where people exit buses, because you tap on a bus but you don't tap off," explains Sager Weinstein.
"What we can do is stitch together the taps and where the bus is to understand how customers are travelling across our public transport network."
This provides a fuller picture of service usage and the areas in which demand is higher and lower to make the end-to-end journey smoother.
They can then group bus routes at a particular popular stop, or place a bus stop nearer a specific tube entrance.
"You can think of that at a very granular level, and really lay out the design of your road network so that it affects where people go, or just redesign a bus network," says Sager Weinstein.
TfL strips out the personal information of customers, and focuses on the journey of a card. By understanding the time they are first used in the morning, the stations they pass and the intensity of the day's travel, they can understand the different types of users, and optimise services to suit their needs.
Serving TfL user segments
Tourists will stop at stations near major attractions until they tap out for the last time at a transport hub such as Heathrow or King's Cross.
They will also take routes that locals often avoid. If a visitor was travelling to Covent Garden for example, they would likely leave the tube at the station of that name, while a Londoner may prefer to get out at the similarly close but less crowded Leicester Square.
Locals also follow different transport patterns, leaving earlier in the morning and following a regular route to their work and back home.
The different segments of passengers also tend to prefer different products. Regular commuters will more likely use a season ticket, while an occasional user normally prefers an Oyster or contactless card.
"That all helps us understand how our customers are travelling, and it helps us provide a better service by thinking about station layouts and messaging," says Sager Weinstein.
"If we have a very stable commuter station, then people will see messages in a different way than at a station that someone's passing through once, but isn't somewhere people regularly come back to. We can use that to think about how we get messages about what's happening on the network and servicing and giving information out to customers."
Journeys through the station
In December 2017, TfL completed a four-week pilot that used depersonalised Wi-Fi data to understand how people move across tube stations. This showed that collecting Wi-Fi data could help provide responsive customer information at specific times of the day on individual lines, platforms and trains.
"With our ticketing information we know at entry and exit from the tube network where people are going but we don't really understand within parts of big complex stations how people move around.
"We don't know how people interchange within the network because you don't have to tap your Oyster card or your contactless payment card. So we did a trial looking at what information we could take from depersonalised connection probe requests from phones connecting to Wi-Fi."
The results showed how crowds changed on trains and then beyond the gateline within periods of time as short as five minutes. They could see how corridors would become congested, and the paths passengers took between points in the station.
This can help them optimise the layout of the station and understand how crowded a train is likely to be, and the best route a passenger could take to find it.
She now hopes to turn the trial into a proof of concept and then build that into a functioning service.
Open data at TfL
TfL is known as a strong proponent of open data, which it offers to developers through its Unified API. TfL's open data is estimated by Deloitte to add up to £130 million of annual economic benefits and savings to London's economy.
This data could offer a vast range of uses for the endless range of businesses and public services affected by public transport.
Third parties can use it to develop their own products or analysis, such as using travel information to help customers save time and collision data to reduce road accidents.
"It's open to anybody to subscribe, and that allows our reach to be even greater," explains Sager Weinstein. "It also of course helps the developer community because it gives them a great feed of data to use."