Stock Price Prediction Using LSTM Model Of Machine Learning And Evaluation Of Its Prediction Accuracy With Reference To Nifty50 And Sensex

The study presents a price prediction of stock framework utilizing machine learning, specifically focusing on (Long Short-Term Memory) as the primary technique. LSTM is highlighted for its capability to capture nuances in stock price behaviour over the specified period. Additionally, the Study propose a machine learning model is based on regularization approach for stock price predictions that aiming to enhance prediction accuracy. Creating a reliable and precise system for analyzing and predicting the future stock price or market trend of Nifty 50 and SENSEX by utilizing the resources such as market indicators, historical stock price data, and other external factors. This aids traders, Investors, and fiscal institutions to make better decisions about purchasing or to decide whether the stock is underpriced or overpriced to buy or sell the stock. In addition, it provides insightful knowledge to financial sector shareholders, the successful development of this system can enhance predictive analytics in the pitch of the stock market estimate. By tackling the system seeks to transform financial market decisions making procedure that majorly enhances performance of the stocks that they own and that reduces the risk of the particular stocks. The dataset sourced from Stock Finance comprises around 900,000 records detailing stock prices and relevant data. Each records include information such as date, symbol, open price, close price, low price, high price, and volume. Solely data from a single company were considered for analysis. Initial data preprocessing involved transforming the CSV-formatted data into Panda’s data frame in Python. Subsequently, data normalization was executed using the Sk-learn library in Python to ensure uniformity in scale across features. The dataset is alienated into 2 sets such as exercise and testing sets, with $50 \%$ allocated for experimentation purposes. The study explores two primary architectures: Deterioration based Model and LSTM. While Regression-based Model employs linear functions for predicting the values continuously, LSTM stances out for its ability to discern changing trends, as evidenced by the results. Consequently, LSTM has identified as the optimal model within the proposed methodology.