Personal positioning and location inference (II)

Abstract

The proliferation of the mobile technologies and high speed internet connection in recent years

has led to an exponential increase in the generation and storage of data. These generated datasets

are often very large in volume and as a result, manual methods of data analysis are rendered

ineffective to extract any useful or accurately identify any patterns of interest. As such, an

emerging field of data mining is being developed in computer science in an attempt to transform

these raw data into useful and understandable patterns.

This project attempts to use data mining techniques implemented in the Java language to identify

patterns of interest from a dataset of GPS coordinates corresponding to a person’s movement

gathered over a period of time. Using techniques such as cluster analysis, the project attempts to

identify locales that are of significance to the user by using various criteria such as the frequency

of which the person returns to the location as well as the cumulative amount of time that the user

spends at a particular location. Further to the abovementioned, the project attempts to identify

patterns of movement by the user and ultimately establish routes or paths that link the significant

locales to one another.

In addition, a visual representation of the results is also generated using a map overlay in a

Google Maps application. The overlay highlights points on the map that have been identified as

significant locales as well as paths linking these identified locales.