“I was very pleased that even the first results were so plausible. That’s rare!” says Karen Cabos. The economist and professor for international management at the Technical University (TH) Lübeck is working with her colleague and Professor Thomas Romeyke on a Big Data project for the Lübeck Business Development Agency. The task: record, process, analyze millions of measurements from sensors in the city and, in a final step, derive forecasts from them. As part of the “City Laboratory for Germany: Vacancies and Settlement” project, the city has been measuring pedestrian frequency at ten locations in the city center since June 17, 2021 using laser-based sensors. Completely without cameras. “With the laser counters, we can determine movements and thus record the frequency of pedestrians. Without taking photos or videos,” explains Cabos.
The Treasurer
“The number of pedestrians measured by the sensors is what we are trying to explain with other data. On the basis of the explanation, we then hope to find adjustments that can be used to influence the flow of people through the city in a targeted manner in the future,” adds Professor Thomas Romeyke. “That’s why we need as much other data as possible in addition to the flow of pedestrians, which can be part of the motivation to walk past one of the sensors right now.” One obvious example is the business opening hours – a very important parameter. While these are relatively easy to take into account, the procurement of other, very detailed data, for example on the current occupancy of parking spaces and multi-storey car parks and local public transport, is more complex: “This data is structured very differently and therefore has to be made compatible with one another in a relatively complex process “, Professor Romeyke describes this part of the project. “Only then can the research work begin,” adds the business IT specialist.
The detective
From the data set prepared by her colleague, Professor Karen Cabos derives a forecast for the number of visitors. Until then, Cabos, like a detective, has to make guesses about what factors might be affecting footfall. This includes, for example, days of the week, times and school holidays, but also weather conditions, parking space utilization, bus frequencies and special factors. After that it’s “trial and error”. “I use several iteration steps to first analyze the explanatory contributions of thematic variable groups such as shop opening hours and finally to integrate them into an optimal forecast model. In the forecasting model, controllable variables – like store opening times – have to be combined with non-controllable ones – like the weather,” explains Karen Cabos. “What it shows: the data is obviously good, since the first results are all plausible.”
The parking lot effect
“What really surprised me was the importance of parking space utilization in the summer months and in December,” says the professor. The utilization of parking lots suggests that many tourists depend on them. “From an ecological point of view, the ratio of cars per person is worse here than in the other months,” Cabos sums up. In addition, there is a strong influence of shop opening and office hours on the number of pedestrians. “The results are quite robust. Our treasure trove of data is still heavily influenced by the corona pandemic. The longer we can collect the data and make comparisons, the better we can predict and draw conclusions about what could help stores and businesses. So the project has to go on,” appealed the professor.
About the City Laboratories project: https://www.stadtlabore-deutschland.de/city/lubeck/