Stock prediction using sentiment analysis on tweets

Investing in stocks has been a topic of interest of many due to its high returns compared to other types of investment such as bonds and savings account. However, investing does not come without risks, and the possibility of incurring financial losses is a significant deterrent for many. The unpredi...

Full description

Saved in:
Bibliographic Details
Main Author: Wan, Qian
Other Authors: Erik Cambria
Format: Final Year Project
Language:English
Published: Nanyang Technological University 2023
Subjects:
Online Access:https://hdl.handle.net/10356/166733
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Nanyang Technological University
Language: English
id sg-ntu-dr.10356-166733
record_format dspace
spelling sg-ntu-dr.10356-1667332023-05-12T15:36:24Z Stock prediction using sentiment analysis on tweets Wan, Qian Erik Cambria School of Computer Science and Engineering cambria@ntu.edu.sg Engineering::Computer science and engineering Investing in stocks has been a topic of interest of many due to its high returns compared to other types of investment such as bonds and savings account. However, investing does not come without risks, and the possibility of incurring financial losses is a significant deterrent for many. The unpredictable nature of potential losses acts as a barrier to entry for many who are not willing to take such risks. Recent studies however have shown that psychological factors such as the public sentiment towards policy changes, new investments, or natural disasters have a significant impact on the stock market's behavior and can be used to predict the price changes of the stock. This study aims to expand on these studies by using 3 natural language processing methods such as Valence Aware Dictionary for Sentiment Reasoning (VADER), Naïve Bayes Classifier, and Bidirectional Encoder Representation from Transformers (BERT). Labelled stock data from Kaggle and self labelled TSLA tweets would be passed through these models which will classify tweets into 3 categories: positive, neutral, and negative. The model that performed the best, BERT, would be used to label a set of unseen TSLA tweets across 6 months and classifying them into their specific categories. After which, Random Forest Classifier would be used to determine if the stock bot should invest or not based on the sentiments predicted using BERT and will be compared against a bot that does not follow BERT’s predictions. Bachelor of Engineering (Computer Engineering) 2023-05-11T12:55:03Z 2023-05-11T12:55:03Z 2023 Final Year Project (FYP) Wan, Q. (2023). Stock prediction using sentiment analysis on tweets. Final Year Project (FYP), Nanyang Technological University, Singapore. https://hdl.handle.net/10356/166733 https://hdl.handle.net/10356/166733 en application/pdf Nanyang Technological University
institution Nanyang Technological University
building NTU Library
continent Asia
country Singapore
Singapore
content_provider NTU Library
collection DR-NTU
language English
topic Engineering::Computer science and engineering
spellingShingle Engineering::Computer science and engineering
Wan, Qian
Stock prediction using sentiment analysis on tweets
description Investing in stocks has been a topic of interest of many due to its high returns compared to other types of investment such as bonds and savings account. However, investing does not come without risks, and the possibility of incurring financial losses is a significant deterrent for many. The unpredictable nature of potential losses acts as a barrier to entry for many who are not willing to take such risks. Recent studies however have shown that psychological factors such as the public sentiment towards policy changes, new investments, or natural disasters have a significant impact on the stock market's behavior and can be used to predict the price changes of the stock. This study aims to expand on these studies by using 3 natural language processing methods such as Valence Aware Dictionary for Sentiment Reasoning (VADER), Naïve Bayes Classifier, and Bidirectional Encoder Representation from Transformers (BERT). Labelled stock data from Kaggle and self labelled TSLA tweets would be passed through these models which will classify tweets into 3 categories: positive, neutral, and negative. The model that performed the best, BERT, would be used to label a set of unseen TSLA tweets across 6 months and classifying them into their specific categories. After which, Random Forest Classifier would be used to determine if the stock bot should invest or not based on the sentiments predicted using BERT and will be compared against a bot that does not follow BERT’s predictions.
author2 Erik Cambria
author_facet Erik Cambria
Wan, Qian
format Final Year Project
author Wan, Qian
author_sort Wan, Qian
title Stock prediction using sentiment analysis on tweets
title_short Stock prediction using sentiment analysis on tweets
title_full Stock prediction using sentiment analysis on tweets
title_fullStr Stock prediction using sentiment analysis on tweets
title_full_unstemmed Stock prediction using sentiment analysis on tweets
title_sort stock prediction using sentiment analysis on tweets
publisher Nanyang Technological University
publishDate 2023
url https://hdl.handle.net/10356/166733
_version_ 1770564493738246144