Personal positioning and location inferences (II)

• Main Author: Ng, Krystal Xing Yi.
• Other Authors: Hsu Wen Jing
• Format: Final Year Project
• Language: English
• Published: 2012
• Subjects: DRNTU::Engineering::Computer science and engineering::Computer applications::Social and behavioral sciences
• Online Access: http://hdl.handle.net/10356/50863
• Institution: Nanyang Technological University

In the recent years, there has been increased usage of phones equipped with GPS functions. This trend has brought about the scenario where the cellphone user is able to generate GPS records continuously. From this continuous data, new types of information such as the cellphone user’s travelling patterns can be obtained. By making use of the individual’s travelling pattern over time, there is the likelihood of carrying out predictions of the future locations that the individual may travel to. This purpose of this project is to develop a program that is able to extract useful information from raw GPS records for the aforementioned travelling patterns and use that information to carry out inferences as to the person’s future destinations. The report focuses on predictions for Personal Positioning, which is specific to each individual and not on human mobility as a whole. Various filtering and clustering methods are explored along the way to find the most suitable ones to clean the data and carry out data mining to gather useful statistical information from the person’s past travelling patterns, which is used in the prediction methods. Three predictions methods (0th, 1st and 2nd Order Prediction) using different basis of information for prediction are developed, tested, and evaluated against each other to see which is the most accurate in predicting the individual’s next destination. All three methods make use of the Markov Assumption and Models, which are described in detail. The report also discusses the benefits and drawbacks of a 3rd Order Prediction method, and explains why it would not be a good improvement upon the 2nd Order Prediction method.