How much data is enough to track tourists? The tradeoff between data granularity and storage costs

Dissertation presented as the partial requirement for obtaining a Master's degree in Information Management, specialization in Knowledge Management and Business Intelligence In the increasingly technology-dependent world, data is one of the key strategic resources for organizations. Often, the challenge that many decision-makers face is to determine which data and how much to collect, and what needs to be kept in their data storage. The challenge is to preserve enough information to inform decisions but doing so without overly high costs of storage and data processing cost. In this thesis, this challenge is studied in the context of a collection of mobile signaling data for studying tourists’ behavioral patterns. Given the number of mobile phones in use, and frequency of their interaction with network infrastructure and location reporting, mobile data sets represent a rich source of information for mobility studies. The objective of this research is to analyze to what extent can individual trajectories be reconstructed if only a fraction of the original location data is preserved, providing insights about the tradeoff between the volume of data available and the accuracy of reconstructed paths. To achieve this, a signaling data of 277,093 anonymized foreign travelers is sampled with different sampling rates, and the full trajectories are reconstructed, using the last seen, linear, and cubic interpolations completion methods. The results of the comparison are discussed from the perspective of data management and implications on the research, especially the results of research with lower time-density mobile phone data.