Prediction and Sentiment Analysis of Stock using Machine Learning

Abstract

Abstract: The stock market is highly volatile and subject to rapid fluctuations and changes, with seemingly insignificant news or rumors potentially impacting the value of a stock. This project aims to assist retail investors in navigating the complex and volatile modern stock market by developing machine learning models for predicting stock market trends and identifying promising stocks. The solution involves components to facilitate data collection through sitemap spider, data processing through coreference resolution technique, model training using random forest, and sentiment analysis for the stock sentiment. The results of sentiment analysis are merged with technical indicators data to create the dataset for training. The proposed system has shown results with 72% accuracy in assisting investment decision making. The project's applications extend to news analysis, investment decision-making, risk management and trading strategies.