What Are Good Situations for Running? A Machine Learning Study Using Mobile and Geographical Data Running is a popular form of physical activity.Personal, social and ecological determinants influence the commitment of the individual. In order to get an insight into the relationship between the running behavior and external situations for different types of user, we have carried out a comprehensive data mining study on large records. We have 4 years historical running data (collected by a mobile practice application of over 10,000 participants) combined with weather, topographic and demographic records. For the analysis of the data we introduce the weighted frequent item mining.In this way, we capture temporal and environmental situations, which are often associated with different mileage. The results show that certain temporal and environmental situations (hour per day, day a week, temperature, distance to residential areas and population density) influence the mileage of the user more than other situational features.Hierarchical agglomerative clustering on the running data is used to divide runners into two clusters (with persistent and less continuous running behavior). We have compared the two groups of runners and found that runners react with less continuous behavior more sensitive to environmental situations (especially different weather and location-related features such as temperature, weather type, distance to the nearest park) as a normal runner.Other analyzes focused on the situation features for the less sustainable runners. The results show that certain characteristic values correspond to a better or worse running track. It was not only examined the influence of individual features, but also the interaction between characteristics.
Our results provide important empirical evidence that the role of external situations in the running behavior of persons from the analysis of the combined historical records can be derived.This opens up a great potential to target these situations in support of people with less sustainable behavior. The whole work can be found under https://www.frontiersin.org/articles/10.3389/full.220.536370/full