Published Paper Details:

The study introduces the MS-SSA-LSTM model, which amalgamates multi-source data, sentiment analysis, swarm intelligence algorithms, and deep learning methodologies to improve stock price forecasting. This model integrates sentiment analysis derived from East Money forum postings, establishing a distinctive sentiment lexicon and computing a sentiment index. This provides significant insights into the impact of market sentiment on stock prices. The Sparrow Search Algorithm (SSA) is employed to refine LSTM hyperparameters, enhancing prediction accuracy. oExperimental findings demonstrate the MS-SSA-LSTM model's exceptional performance. It is an essential instrument for precise stock price forecasting. Designed for India's unpredictable financial market, the model specialises in short-term stock price forecasts, providing insights for agile decision-making by investors. A hybrid LSTM+GRU model was developed for stock sentiment classification. A comprehensive ensemble approach was implemented, utilising a Voting Classifier (AdaBoost + RandomForest) for sentiment analysis and a Voting Regressor (LinearRegression + RandomForestRegressor + KNeighborsRegressor) for stock price forecasting. These ensembles were smoothly merged with existing models (MLP, CNN, LSTM, MS-LSTM, MS-SSA-LSTM), thereby improving overall predictive performance. A user-friendly Flask framework with SQLite support was designed to enhance user engagement and testing, simplifying the signup, signin, and model assessment processes.