University of Twente Student Theses


Sentiment Analysis on Social Media using Machine Learning-Based Approach

Agustini, Try (2021) Sentiment Analysis on Social Media using Machine Learning-Based Approach.

[img] PDF
Abstract:This study investigates the opinions and thoughts written by social media users about a particular product then determines the relationship between its underlying sentiments and a financial indicator of the said products. The two main topics selected as the case study are Bitcoin cryptocurrency and a ridesharing application service named Uber, obtained from Twitter posts within a time period of the year 2019. The goal is mainly to answer three main research questions. The first one is to identify which machine learning algorithm has the best performance in the classification task of sentiments of Twitter data. The second is to find out to what extent the sentiments on social media can affect the financial performance of a product. Lastly, we will observe how the outcomes of this correlation based on the machine learning method compare to the lexicon-based approach. This study shows that boosting algorithm has performed the best among the other machine learning techniques. It is also evident that positive sentiments have a more significant relationship with Bitcoin price, whereas negative sentiments have more influence on Uber price. Finally, these findings are consistent with the experiment result of the lexicon-based approach, which means the study proposed with the machine learning approach is proven to be reliable.
Item Type:Essay (Master)
Faculty:EEMCS: Electrical Engineering, Mathematics and Computer Science
Subject:54 computer science
Programme:Business Information Technology MSc (60025)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page