STOCK PRICE PREDICTION USING MACHINE LEARNING AND SENTIMENT ANALYSIS

Abstract : The stock market is widely regarded as one of the most dynamic and unpredictable sectors, where changes can occur rapidly due to a myriad of factors such as market sentiment, economic events, and geopolitical influences. Unlike other traditional markets, the stock market is subject to frequent fluctuations, which makes it challenging for investors and traders to accurately predict trends. In recent years, advancements in technology have opened up new avenues for analyzing and predicting stock market movements. This paper explores several methods that have been employed to dynamically learn and adapt to the market’s ever-changing nature, with a particular focus on prediction models and sentiment analysis. In this study, we applied three different models to forecast stock prices and trends, each representing a distinct approach in analyzing the data. The first model uses traditional time-series forecasting techniques such as the Autoregressive Integrated Moving Average (ARIMA) model, which is particularly well-suited for modeling and predicting future stock prices based on historical data. The second model incorporates machine learning algorithms, including decision trees and support vector machines, to learn complex patterns from the data and make predictions based on these patterns. The third model incorporates natural language processing (NLP) to conduct sentiment analysis on social media, particularly tweets, related to the company or stock in question. This sentiment analysis aims to capture the public’s emotions and opinions, which can significantly influence stock prices. Sentiment analysis plays a critical role in understanding how news, events, or public opinion can impact the stock market. Twitter, in particular, has become a popular source of real-time information, with millions of tweets generated daily, often reflecting the general sentiment of investors and the public. By analyzing these tweets, we can gauge the mood of investors and understand how certain events or market movements influence stock sentiment. In this paper, sentiment analysis was conducted using a combination of text mining and machine learning techniques to extract valuable insights from tweets and correlate them with stock price movements. The performance of each model was evaluated based on its accuracy in predicting stock price trends. Among the three models, the ARIMA model performed the best, providing the most accurate predictions for every stock tested. Its ability to model stock price movements based on historical data with minimal error makes it a valuable tool for short-term forecasting. The machine learning models, while effective in certain contexts, faced challenges in generalizing to the market’s inherent unpredictability. The sentiment analysis approach, while insightful, did not always correlate strongly with market movements, as the stock market is influenced by numerous factors beyond public sentiment. The results from this study provide valuable insights into the random and often erratic nature of stock market fluctuations. By using ARIMA alongside sentiment analysis, investors can gain a better understanding of potential trends and make more informed decisions when trading stocks. The ARIMA model, in particular, stands out as the most reliable for stock price prediction, offering a method for investors to reduce risk and make decisions based on historical trends and patterns. This paper, therefore, provides not only a comparison of predictive models but also a new approach for investors to evaluate stock market opportunities and manage their portfolios effectively. The findings have implications for both retail and institutional investors, offering them tools to navigate the complex world of stock market prediction with a higher degree of confidence.

Keyword Sentiment Analysis, ARIMA, LSTM, Linear Regression, Naïve Bayes, Stock Market Prediction , Tweets