Degree

Bachelor of Science (Computer Science)

Department

Department of Computer Science

School

School of Mathematics and Computer Science (SMCS)

Advisor

Tasbiha Fatima, Lecturer, Department of Computer Science, Institute of Business Administration, Karachi

Co-Advisor

Hunzala Mushtaq, Software Engineer, Folio3

Keywords

Stock Price Prediction, Machine Learning, Sentiment Analysis, Time-Series Forecasting, Natural Language Processing (NLP)

Abstract

The Stock Price Forecaster is an advanced AI-driven platform developed to predict stock prices by synthesizing quantitative historical market data and qualitative real-time news sentiment analysis. Leveraging the power of Bidirectional Long Short-Term Memory (LSTM) neural networks, the system models temporal dependencies in financial time series data. It integrates dynamic market sentiment extracted through natural language processing techniques. By delivering multi-horizon forecasts—specifically for 1-day, 15-day, and 30-day intervals—this platform aims to provide users with actionable and timely predictions that improve upon traditional models limited to historical price data. The inclusion of news sentiment enables the model to capture the emotional and psychological market factors that influence stock volatility and price movements. An interactive web-based dashboard serves as the user interface, providing visualisation of forecasted trends, sentiment overlays, and downloadable prediction data. This design makes complex forecasting accessible to a broad spectrum of users, ranging from retail investors to professional analysts. This report outlines the methodologies, experimental settings, results, and future directions of the Stock Price Forecaster, demonstrating its effectiveness and potential impact on data-driven investment-making.

Tools and Technologies Used

  • Programming Languages: Python
  • Machine Learning Frameworks: TensorFlow, Keras
  • Natural Language Processing: TextBlob, NLTK (for sentiment analysis)
  • Data Fetching APIS: Yahoo Finance (via yfinance), Newsapi
  • Web Framework: Flask (for backend and serving HTML pages)
  • Frontend: HTML, CSS, JavaScript (including Chart.js for interactive charts)
  • Model Tuning: Keras Tuner for hyperparameter optimisation
  • Visualisation: Matplotlib (for training plots), Chart.js (for web charts)
  • Data Processing: Pandas, NumPy
  • Version Control / Collaboration: Git

Methodology

The development followed an agile methodology with iterative sprints involving continuous feedback from academic and industry mentors. The project was divided into modules: data gathering, model development using LSTM, sentiment integration, and web interface design. Real-time news and sentiment streams were linked to the prediction engine, allowing the system to evolve in sync with market behaviour. Each sprint culminated in testing and refinement to ensure usability, performance, and relevance.

Document Type

Restricted Access

Submission Type

BSCS Final Year Project

Download

Share

COinS