Application of Supervised Machine Learning Techniques for Predicting Nigeria’s Gross Domestic Product

Abstract

Gross Domestic Product (GDP) is a vital indicator for assessing a country’s economic growth. This study applies linear regression, a machine learning technique, to accurately predict GDP using independent variables such as private consumption and gross private investment. Past prediction models faced challenges due to inappropriate model selection and data gaps, resulting in inaccurate forecasts. This research aims to overcome these limitations by developing and testing a linear regression model and comparing its performance with previous models, including the K-Nearest Neighbors (KNN) algorithm. The model development process involved data acquisition, cleaning, transformation, splitting, training, and performance evaluation using Mean Squared Error (MSE). A dataset of 39 complete observations was used. Comparative results indicate that the linear regression model achieved a prediction accuracy of 98.2%, outperforming KNN (89%) and other regression methods such as Ordinary Least Squares (88.9%), Ridge Regression (88.5%), and Principal Component Regression (87.3%). The findings, supported by visual comparisons of actual and predicted GDP values, demonstrate that linear regression offers superior predictive performance for GDP estimation. The model's high accuracy and low error rate highlight its reliability and effectiveness in economic forecasting.