Assessment of Spatial Trajectories Repeatability Using the Hausdorff’s Distance

• Activity Number: 219 - Contributed Poster Presentations: Section on Statistical Graphics
• Type: Contributed
• Date/Time: Tuesday, August 4, 2020 : 10:00 AM to 2:00 PM
• Sponsor: Section on Statistical Graphics
• Author(s): Josep Carrasco* and Carola Sanpera
• Companies: University of Barcelona and University of Barcelona
• Keywords: Repetability; intraclass correlation; Hausdorff's distance; Trajectories

Abstract

In recent years, the availability of georeferenced data has allowed to obtain information about animal movement. Flight trajectories of 36 gulls were obtained through 18 days in the Ebro Delta Natural Park (NE Spain). One of the goals of this research was to assess the repeatability of the gull’s trajectories in order to establish if there are different subject’s strategies or if, on the contrary, the flights are at random. Repeatability is usually assessed by the intraclass correlation coefficient (ICC) that is defined as the proportion of the total variance accounted for between-subjects variability. The ICC is usually estimated by variance components of linear mixed models (LMM). In this context the data design involves taking repeated measures on every subject. In our example the repeated measures are the different trajectories across days. However, the ICC for the trajectories cannot be estimated using the common approach based on LMM because each trajectory consist of sets of spatio-temporal measures rather than single readings. Here we will propose a novel approach to estimate the ICC for trajectories based on Hausdorff’s distance.