Application of Machine Learning and Deep Learning in Stock Price Prediction: A Comparative Study Using Dangote Cement Plc Dataset from the Nigerian Stock Exchange
Article PDF

Keywords

LSTM; Random Forest; SVM; stock price prediction; Dangote Cement; Nigerian Stock Exchange; machine learning; deep learning; financial forecasting; institutional-based research

How to Cite

Salisu, A. (2026). Application of Machine Learning and Deep Learning in Stock Price Prediction: A Comparative Study Using Dangote Cement Plc Dataset from the Nigerian Stock Exchange. Global Journal of Artificial Intelligence and Technology Development , 4(1), 119-128. https://doi.org/10.46654/5da6a248

Abstract

This study applies three machine learning and deep learning algorithms — Long Short-Term Memory (LSTM) networks, Random Forest (RF), and Support Vector Machines (SVM) — to predict the daily stock prices of Dangote Cement Plc (DANGCEM) on the Nigerian Stock Exchange, using a decade-long dataset of 2,750 daily trading records from January 2013 to November 2023. The study was conducted as institutional-based research at Jigawa State College of Education and Legal Studies, Ringim, aiming simultaneously to generate a locally contextualised financial prediction model and to build institutional capacity in financial data science. Data preprocessing included missing value forward-fill, Min-Max normalisation, feature engineering (lag prices, moving averages, daily returns), and an 80/20 chronological train-test split (n_train = 2,200; n_test = 550). LSTM achieved the highest performance across all seven evaluation metrics: accuracy = 87.4%, precision = 86.2%, recall = 88.1%, F1-score = 87.1%, RMSE = ₦12.34, MAE = ₦8.62, and AUC = 0.924. Random Forest achieved accuracy = 82.6% and AUC = 0.882; SVM achieved accuracy = 79.3% and AUC = 0.851. Feature importance analysis confirmed that the lag t-1 closing price (permutation importance = 0.328) is the strongest predictor, followed by the lag t-5 price (0.184) and the 5-day moving average (0.138). The Institutional Financial Analytics Research Framework (IFARF) is proposed as a replicable template for financial ML research within Nigerian Colleges of Education. The study identifies real-time data integration, model interpretability, and external shock robustness as priority directions for future research.

Article PDF
Creative Commons License

This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.

Copyright (c) 2026 Global Journal of Artificial Intelligence and Technology Development

286 234