Mining Twitter Data for Resource Usage Prediction

Abstract

This thesis investigates the predictability of Twitter traffic for topic-related websites’ resource requirements by developing and implementing a data mining methodology. The new traffic correlation mining process is able to extract traffic surges and develop potential predictive mining and correlation techniques between Twitter and the corresponding forum. Thorough testing of this data mining methodology has been performed, and the results show that using Twitter data to predict imminent resource demands is a fruitful area of research. The findings in this thesis confirm the potential of utilizing the significant public interests expressed in Twitter data as a resource usage prediction tool for relevant websites.