The urban data analyst Anna Lvova discusses, why people’s movement in the city should be monitored.
The transport analytics is one of the most difficult and undisclosed subjects in urbanism. It involves a lot of calculations, metrics and mathematical models, and that is why it’s being performed by domain specialists by request of field experts. That’s why it doesn’t seem interesting to the public. However, the studies, targeted at both professional transporters and ordinary people, have started to appear more often lately and tend to be public. «Home — work, work — home» by Yandex and Movement by Uber are among them. How do the city government, scholars and independent researchers feel about these kinds of studies? Strelka Magazine asked Strelka’s alumni, the urban data analyst in Habidatum Anna Lvova to talk about the advantages and limits these researches have and why there will be more and more of such works in the future.
HOW THE TRANSPORT IS BEING MONITORED IN MOSCOW
In urbanism, the big data, as the title suggests, gives information about a much larger amount of people, than we could ever imagine. The sociological methods — interviews, questionnaires or observations — simply can’t have such coverage. Old quantitative studies take quite long, they are expensive and limited by the territory or time: you can ask all the city residents, what they buy, while completely overlooking the questioning of commuters. Or you can measure pedestrian traffic every two hours and overlook the anomaly within that time interval.
How mobile and active the population is? To answer this question scholars used to refer to the population census, the information about the real estate purchases and the registration of companies. These practices are very uncommon in Russia for obvious reasons: the internal migration that existed in America or Europe was out of the question in the Soviet Union, which is why it wasn’t so thoroughly studied. Although, this situation has advantages. The absence of customary tools makes cities more open for new technology because certain stages of development can be skipped.
«Moscow has one of the most progressive intellectual transportation systems in Europe. The city has two transport models: the static one and the dynamic one. The latter involves plenty of data going to the Command post of Center of the traffic organisation in real time. It shows, what city motion looks like at the specific moment in time.
Generally, there are many parameters that are being and monitored by the transport planners. The analysis of transport demand is usually conducted with so-called origin-destination matrices. The basic model includes a table, which shows the areas of the departures, arrivals and a number of transitions between them. In order to understand the transport availability of a certain area, the isochrone method is implemented: the travel time on any transport within a point or a whole area is being estimated and mapped. We used it, planning the «Magistral» surface transport network. You can measure the rate of the flow — the concentration of people and the load of different sections of the road-street network. For example, you know that there’s always a traffic jam on Tverskaya street at 10 a.m., but you don’t know where all these people are coming from.
The basic model of the origin-destination matric includes a table, which indicates the areas of the departures, arrivals and a number of transitions between them
It is much more difficult to understand, how many people there are and what kind of transport they use. It is still being measured with surveys, but it is very expensive because a large city requires for corresponding selection. Such survey has been conducted in Moscow for three times, the last one — in 2015. What transport researchers are missing is the coordination between various types of data. For instance, the metro or land transport. We don’t see this data in real time, we can’t compare them to the car flow, even though we have an infrastructure for it. Roughly speaking, we now have more data than the understanding of what to do with it. And we also terribly lack good-quality automated data on pedestrian traffic«.
WHO ELSE ANALYSES BIG DATA ASIDE FROM THE SCIENTISTS
The Moscow transport model is being supplied by several data sources. The main one is Center of the traffic organisation’s own sensors and cameras that are set all around the city, and the additional one is housing machinery, buses, parking sensors that register parking violations. The government also made a deal with Uber, «Yandex.Taxi» and Gett taxi services, so now it receives depersonalised data from them. This kind of practice is not common in the world. The American government has tried for a long time to make local taxi services like Uber and Lyft give in their data to the service of the city.
Uber: To everyone or to no one
American Uber refused to cooperate and it has tried to sort things out with the government in the court for a long while. So the company found a third way — sharing their data with everyone. It launched Movement , a service, which shows the statistics of taxi traffic. The service also demonstrates, how the road workload changes with time. You can pick start and finish areas for Movement to indicate, how long the trip will take you: right now or on average, during a certain time period. Now you need to request access to use the service, but it is promised to become fully public. For now, there are other projects available. The information that’s presented there is the aforementioned origin-destination matrices, nicely visualised. Also, they have a time factor, which makes that information extremely valuable (especially considering that Uber receives that information for free, unlike transport planners).
«Yandex»: From transport to urban planning
In December 2016 «Yandex» published «Home — work, work — home» study on Moscow citizens’ daily traffic. The quotidian cycle of migration was analysed, using data from «Navigator» and «Yandex.Maps». As a result, Moscow districts were divided into residential, work and mixed. «Yandex» also analysed the travel time: when people start the trip, how long it lasts and how traffic jams influence that. That study is valuable not only for transporters but also for urban planners, as it tells a lot about citizens’ behavioural features and the typology of the district. The social geographer Olga Vendina gave a comment on the results of this study and highlighted the most interesting districts.
«Despite the importance of such factors as location and distance between home and work, there are other even more influential factors for the drivers: work schedule (night, shift, «a day after three», freelance, schedule limitations, working from home, standard schedule); status, work position and the nature of work; social status, including the responsibilities of mother and housekeeper, lifestyle and, finally, the advancement of the alternative possibilities of moving from point A to point B. Thus, the daily car traffic rate graph shows some significant differences between citizens of Zvenigorod, Maryino district, Akademichesky district, and Tverskoy district.
The typical elements for Maryino are: relatively smooth curve line of the traffic throughout the day, the absence of sharp peaks and a noticeably higher amount of trips «home» early in the morning and in the night. That means that most of the residents work in shifts, not by the standard schedule (from 9 a.m. to 6 p.m.). There’s also another evident conclusion. Despite its grand size, Maryino’s ground public transportation and metro systems are developed much worse than in any other Moscow district. It’s a kind of «urban island». This situation makes people travel everywhere by car: to the school, to the hospital, to the shop, to a friends’ house. That is why the departure and destination points that are being entered in the navigator systematically can’t be clearly defined as home or work.
There are two flows that are, apparently, being formed in Zvenigorod. People from the first flow just go about their business (probably, to work) in the morning to get there in time by 9-10 a.m. (about 30-35% of the car drivers), and people from the second flow — to reach the destination after 10 a.m. That district has the most arched curve line in the evening period. That means that there’s a large group of people, who, first of all, prefers coming to the office a bit later (that is hardly possible with the manufacturing companies) and leaves later as well. Second, it can be people, who spend their free time in the evening outside of, presumably, their homes. Zvenigorod residents seem more dependent on the cycle of residents’ activities. To be more precise, that cyclicity is less individual and more group-based, which is noticeable due to groups’ «standard» behaviour.
Tverskoy district, on the other hand, is less dependent on the daily schedule and the necessity of actions. The fact that about 40% of car owners drive from 9 a.m. to 12 p.m., shows that work is not the only purpose of car trips. There are many alternative purposes: from «take to school — take from school» to rides to the gym. It is quite possible that it might be caused by a higher level of population’s automobilization, wherein having two cars in one family is not uncommon.
Akademichesky district can be characterised as intermediate between Tverskoy district and Maryino. It is a district with a more standard working day cycle, where the main activity starts at 9-10 a.m., the peak of business meetings falls on post-lunch time (4-7 p.m.), and people rush back home after 7 p.m. This is probably the middle class with its own standards and lifestyle. It may seem weird, but the daily schedule of Akademichesky district’s residents looks more monotonous than Zvenigorod citizens’.
A bit later «Yandex» released its other study, regarding the use of public transport. It measured the popularity of ground transport stops in various districts of Moscow. The number of people, using «Yandex. Transport» app in close proximity to the stop was the main criteria.
Olga Vendina: «That study is less informative, but it also conjures up many interpretations. First, it is obvious that a number of stops is bigger in the areas where the density of the road-street network is higher and where its connectivity is provided better. It gives an opportunity to organise the ground transportation route network, which makes it more appealing to use. It also saves time and reduces a number of transitions. No other area of Moscow can compete with the south-west on that account, which is why that part is the most visible on the map. You can notice the districts that already have their own developed bus route networks: Izmaylovo, Yasenevo, Solntsevo, Orekhovo-Borisovo; and also the districts that don’t have such network yet: Maryino, Lyublino, Metrogorodok. There can be many reasons for that, but it is difficult to find the ones that would explain it fully».
POSSIBILITIES OF BIG DATA
Big data is a hot topic, and it starts being actively used by the city administration. The mayors and managers say more and more often that they learned to make decisions, based on objective information —, not experience or intuition. Still, it’s important to realise that this «objective» approach is not a universal solution to all the problems, and it has a lot of limitations that we should know about.
First of all, such data presents the selection of the entire population. And it doesn’t necessarily cover all the citizens. For example, the «Yandex» study showed that all the central districts are work only. It looks like no one lives there. But the fact is that people who live in the center simply may not use personal transport, therefore they are not displayed in the research. Strictly speaking, that study is not about «where Moscow citizens live and work», but about «where Moscow drivers, who use „Yandex“ services and keep their home and work addresses there, live and work».
Even though Uber made a revolution on the American market by making taxi more available — still, its regular user is probably richer than a regular citizen that uses public transport. There is also another problem, concerning age and even class representativeness. Uber is a mobile app that is popular among young, technologically advanced people. The problem of selection can be solved by using different kinds of information or different filters for the same source. «Yandex» registered the time, when people go from home to work. For one-third of the observed, it is the period between 7 and 9 a.m. Now it would be interesting to see all the trips within that period. Most of the «Navigator» users keep their home and work addresses in the app. So we can assume that most of the trips during this period are made to get to work. That would allow us to get information about a much larger amount of people. However, some parents, taking their children to school, or couriers could get into the selection. That would make data messier.
OTHER WAYS OF RESEARCHING URBAN MOBILITY
More and more ways of locating people are being discovered as the digital technology develops. The phone itself provides quite a few options to track its owner’s location. First of all, the mobile operators have that information. Any SIM card could be detected even back in the day when we used mobile phones with polyphony and an integrated flashlight. That source still has one of the largest coverages, because everyone has a phone and 80% of the communication market belongs to three companies.
The brightest example of this data’s usage is the « Archaeology of the periphery» research, conducted by Thomson Reuters, Matrioshka and Megafon for Moscow Urban Forum in 2013. They analysed Moscow citizens’ traffic on a working day: where they come from, where they go and what highways they cross. It turned out that two-thirds of Big Moscow residents don’t go outside their districts on a working week, and one-fifth of the morning trips to the centre is nothing but overruns. That means that the citizen doesn’t have to go to the centre, but that is the only way to make a transition or to get to another radial highway. It was that research that debunked a myth about Moscow being a super mobile city and it showcased the deficit of chord highways.
The second source that hasn’t proved itself yet, is operating systems’ manufacturers. For example, all the iOS and Android devices record your location on default — that way you can find your lost phone. These two platforms own 98% of the world mobile communication market. Google and Apple are not ready to share that information yet, but Uber’s example can inspire the IT giants.
Third, user’s location is available to many mobile apps. The location services are often used in the navigational products, such as TomTom or Here, running apps like Nike+ and lots of others. TomTom is the one that deals with traffic analytics: every year it presents a rating of cities with the biggest traffic jams with Moscow leading it until 2013. Here provides that data to city government: it collaborates with the transport departments of several American states. Other companies usually use data to optimise their apps and also for the purposes of individual analytics, for instance, for a user to be able to see the statistics of all of his or her runs.
It is interesting that the information concerning user’s location is available to a much bigger amount of apps than we think. Try going to the «Location» menu in your phone settings — you’ll be surprised to see how many services — from music apps to games — know, where you are. The more popular the app is, the more interesting it would be to work with its data. Thus, PokemonGo app can have more information about Moscow pedestrian routes than the transport planners.
Aside from apps’ tracking, there’s also plenty of open information that people share themselves on Twitter, Instagram, Foursquare or Facebook. It’s one of the most popular and actively developing areas of urbanism. Open data supplies many researchers and allows to learn some unexpected facts.
Commuting and the borders of agglomerations were researched by Habidatum in five Russian cities, based on Instagram. There’s another example — the interactive map of the most photographed places in the world, based on the data from 500px — the website for professional photographers.
Finally, the flows can be analyzed via CCTV cameras on the roads, detectors, installed in the asphalt, smart lanterns and other sensors. And that is only a small part of the list. There’s another developing field, called financial transactions analysis, which is the process, where your credit card tells you where and when you make purchases and, thus, how you move around the city. There are very few studies in that field. And they were mostly conducted for academic purposes till now, in order to demonstrate the relevance of the approach. For now, it’s either the banks are afraid of sharing that information with the analysts or the urban analysts themselves don’t know how to apply it. However, all the data types, commonly used now, went through that stage: from mobile operators to social networks. Uber itself already managed to partner with the researchers from MIT Senseable City Lab and publish a small part of their data concerning average waiting time in different parts of the city. I think that more and more data holders and producers will share it with urbanists. But it will get even more interesting, when all the apps, tracking user’s movements, will become urbanists themselves. So that we will have the researches from Spotify, Pokemon.Go and Meitu instead of analytical reports from the city government.