Analysis of Text Corpuses From Linguistic Aspects Using Python

Abstract

Natural Language Preprocessing (NLP) tasks like sentiment analysis and text classification have gained popularity as techniques to estimate investor sentiment towards a certain stock. This study performs a stock sentiment analysis using textual data from Reddit related to the TSLA stock. It employs a degree of machine and deep learning algorithms to train a classifier model for multi-class classification of the text sentiment, incorporating novel techniques for sentiment reproduction. Additionally, the study investigates the impact of data preprocessing techniques such as undersampling and tokenization on the performance of the sentiment analysis model, providing insights into the effectiveness of these methods in handling imbalanced textual data. The research also explores the influence of machine learning on discerning stock market sentiment, interpreting its ramifications on investor behavior and market dynamics.