Trip Activity Chain Pattern Recognition and Travel Trajectory Data Mining

This study focuses on recognizing the travel modes and activity types in personal travel trajectory. Firstly the paper proposes a concept of trip-activity chain pattern to describe the general form of travel trajectory, and analyzes the structure and features of this pattern and its sub-patterns: trip sub-pattern and activity sub-pattern. Then normalized Euclidean distance measurement method is adopted to decompose the travel trajectory into trip and activity parts. Finally the authors apply RBFNN to solve the pattern recognition problem, which obtains an accuracy of 88.5% of travel mode recognition and 74.4% of activity type recognition.