Time Series Analysis and Stock Price Prediction on Microsoft Dataset

Abstract

As the complexity and volume of financial market data continue to grow, accurately predicting stock prices has become increasingly challenging. Traditional statistical and machine learning approaches often struggle to capture nonlinear patterns and temporal dependencies present in time series data, resulting in limited forecasting accuracy [6], [10]. This paper presents a deep learning-based stock price prediction system using Long Short-Term Memory (LSTM) networks to improve prediction performance on the Microsoft (MSFT) dataset obtained from Yahoo Finance [5]. The system processes historical stock data, converts it into sequential patterns, and learns trends to generate future price predictions using LSTM’s capability to model long-term dependencies [1], [2].

To enhance practical usability, the trained model is deployed using a Flask-based backend, enabling real-time prediction through RESTful API communication. An interactive web interface developed using HTML, CSS, and JavaScript allows users to input data, request predictions, and visualize results through dynamic graphs. The implementation is supported by deep learning frameworks such as TensorFlow, which facilitate efficient model training and deployment [4], [8]. By integrating data science with web technologies, the proposed system provides an efficient, scalable, and user-friendly solution for financial forecasting and decision-making.