Clustering Travel Behavior Time Series Using Topological Data Analysis

• Activity Number: 462 - SAMSI Program on Transportation Statistics
• Type: Topic Contributed
• Date/Time: Wednesday, August 1, 2018 : 8:30 AM to 10:20 AM
• Sponsor: Transportation Statistics Interest Group
• Author(s): Renjie Chen* and nalini ravishanker and Jingyu Zhang and Karthik Konduri
• Companies: and University of Connecticut and University of Connecticut and University of Connecticut
• Keywords: Activity-Travel Patterns; Categorical time series; Clustering; Topological Data Analysis

Abstract

The study of generational cohort effect on activity-travel patterns (ATP) is of interest to transportation professionals. In characterizing ATP, most of the research to date has only focused on univariate characteristics of activities and travel (purpose, timing, mode, accompaniment), also often ignoring temporal dimension (timing and duration). There is a need to characterize ATPs by considering the full set of characteristics, also recognizing the temporal attributes to holistically analyze the impact of generational cohort effect on ATPs. The objective of our research is to analyze multiple waves of national household travel surveys from the last 20 years using topological data analysis (TDA). TDA is an emerging area for the statistical analysis of complex data aimed at summarizing features using homology groups. ATPs are characterized as categorical time series and clustered using TDA. The demographic composition of ATP types are explored to study generational cohort effects while accounting for other factors. TDA can also be used to analyze other complex categorical time series in the transportation domain (e.g. driving behaviors).