Text Analysis for Financial Time Series Data Modelling

📄 Project Abstract

This project aims to explore the field of sentiment analysis of financial news data to attempt to predict and model stock and share prices based on Natural Language Processing techniques. A lot of research of stochastic data has created good models based on past numerical data but analysis on text data to extract sentiment has not been explored to its potential. Using simple techniques such as Bag of words method with some additions and then building upon this adding features such as Part of Speech tagging and sentence context analysis can vastly improve performance for such algorithms. By using classifying techniques in WEKA using Naïve Bayes and Multilayer Perceptron algorithms a solid classifying system can be built to evaluate and even predict performance of these shares.