Comparison of Similarity Measures for Trajectory Clustering - Aviation Use Case

• DOI: 10.24138/jcomss-2022-0116
• Publication Date: 2023-07-03T13:36:55Z
• Authors: Marija Todoric; Toni Mastelic
• Keywords: Hausdorff distance; DBSCAN; Similarity measure; Similarity (geometry); Robustness

Various distance-based clustering algorithms have been reported, but the core component of all them is a similarity or distance measure for classification data. Rather than setting priority to comparison performance different algorithms, it may be worthy analyze influence measures on results algorithms. The main contribution this work comparative study impact 9 similarity-based trajectory using DBSCAN algorithm commercial flight dataset. novelty in exploring robustness with respect parameter. We evaluate accuracy clustering, anomaly detection, algorithmic efficiency, and we determine behavior profile each measure. show that DTW Frechet lead best results, while LCSS Hausdorff Cosine should avoided task.