Mining raw GPS readings for deep profiling of location contexts - part II

• Main Author: Ng, Zheng Kai
• Other Authors: Luo Jun
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2024
• Subjects: Computer and Information Science
• Online Access: https://hdl.handle.net/10356/181106
• Institution: Nanyang Technological University

This project focuses on enhancing location context profiling through the analysis of raw GNSS data, collected in Singapore using the GNSSLogger application on an Android device. Leveraging the richness of the raw satellite data—comprising latitude, longitude, and satellite metadata. This study aims to provide a more detailed and nuanced understanding of the environment in which the data was captured. By integrating the OpenStreetMap API, additional spatial information was extracted to contextualise the locations, offering insights beyond basic coordinates. The raw data underwent rigorous preprocessing, including the removal of invalid coordinates and noise filtering, to ensure accurate analysis. To classify different location contexts, several machine learning techniques were evaluated, including KMeans, Random Forest, DBSCAN. DBSCAN was found to be the most effective, particularly in complex urban settings where traditional classification methods often fall short. The results demonstrated significant potential in using machine learning to refine GNSS-based location profiling, advancing the accuracy and reliability of context-aware systems in dynamic environments. This work lays the foundation for future developments in location-based services, contributing to more intelligent and responsive urban navigation systems.