Exploring Human Mobility: A Time-Informed Approach to Pattern Mining and Sequence Similarity

The surge in the availability of spatial big data has sparked increased interest in researching human mobility patterns. Despite this, discovering human mobility patterns from such spatial big data and assessing the similarity between patterns remains a formidable challenge. This study introduces two novel methods: the Time-Informed pattern mining (TiPam) method for frequent pattern mining and a Time-Aware Longest Common Subsequence (T-LCS) algorithm for assessing similarity between time-conscious sequences. Leveraging these innovative algorithms, our research introduces an analytical framework for analyzing human mobility patterns at both individual and aggregated levels. As a case study, this proposed workflow is applied to examine the daily mobility patterns of voluntary mobile phone users in Kampala, Uganda. The 135 participants are found in four distinct groups labeled with distinct mobility properties for users in each group: "stay-at-home," "unoccupied," "education-oriented," and "work-oriented." The results effectively showcase the efficiency of the framework and the novel techniques employed. The framework's versatility extends to human mobility studies with other forms of data and across various research fields.