Visual analysis to generate and validate geographical heuristics

The popularity of smartphones has opened endless possibilities for researchers. The adoption of these devices by all kind of users has transformed each one of them into a mobile sensor that can measure different types of data. These data can eventually be transmitted via Internet to servers where it can be studied to understand different aspects of the environment and human behavior in ways that were not possible a few years ago. Even with all this information available, the granularity of the data can be coarse, and the amount of it can be scarce, making it difficult to study and evaluate such data. This is especially true for certain geographically-related problems, where conventional algorithms are difficult to design. For these problems it is possible to design heuristics to reach a good approximation, however the methodology to assess the effectiveness of such heuristics is unclear. In this paper we present our work towards a methodology to generate and evaluate such heuristics. This methodology is based on displaying geographical information visually so human evaluators can judge the quality of the heuristic and combining it with the calculation power of computers.