FORECASTING NIGERIA'S POPULATION GROWTH USING DEEP LEARNING TECHNIQUE

Effective governance relies on accurate population data to support resource management, policy-making, and informed decision-making. In Nigeria, census efforts are essential but face challenges such as political bias, ethnic and religious tensions, logistical issues, and technological limitations, resulting in data inaccuracies and highlighting an urgent need for reliable population estimation methods. The research proposes a novel approach to forecasting Nigerian state populations using a Recurrent Neural Network (RNN) with Long Short-Term Memory (LSTM) units. The model was developed in Anaconda IDLE, leveraging historical state-wise population obtained from the Demographic Statistics Bulletin published by the Bureau of Statistics. This bulletin compiles population projections and growth trends derived from official national census figures, administrative records, and demographic surveys conducted periodically by government agencies. The data provides state-wise population counts, ensuring comprehensive coverage of the study region. The collection process involved manual extraction of relevant data points comprising 629 samples, verification against supplementary reports to ensure accuracy, and preprocessing to standardize the data for analysis and forecasting. Model accuracy was evaluated using Mean Absolute Percentage Error (MAPE) recording a 2.97% accuracy and Symmetric Mean Absolute Percentage Error (SMAPE) with 2.90% accuracy. The results indicate a substantial improvement in population forecasting reliability. This approach offers policymakers enhanced confidence in population data, demonstrating the potential of deep learning techniques in addressing complex societal challenges.