A Time series analysis of population growth trends in Bindura town, Zimbabwe

Author

Chitau, Tanyaradzwa

Title

A Time series analysis of population growth trends in Bindura town, Zimbabwe

Abstract

This study investigates Bindura Town’s population dynamics from 1990 to 2023 and produces forecasts for 2025–2028 using two time-series modeling approaches. Drawing on over three decades of demographic data, the analysis applied an ARIMA (2,0,1) model identified via Box Jenkins methodology and selected for optimal performance using AIC and BIC criteria which achieved strong accuracy (MAPE = 5.82%, R² = 0.79). A Long Short-Term Memory (LSTM) neural network, configured with a 12-month lag window and multiple hidden layers, was also implemented to capture nonlinear patterns in the data, though its performance (MAPE = 10.35%, R² = 0.76) was slightly less effective for this case. Both models project modest but consistent growth in Bindura’s population, from approximately 105,822 in 2025 to 108,210 by 2028, with the 25–54 age group particularly among females contributing most significantly to the increase. These results suggest that ARIMA remains a robust tool for municipal forecasting, offering interpretable and statistically reliable projections that can inform evidence-based planning for housing, infrastructure, and public services in a rapidly urbanizing context.

Date

June 2025

Publisher

BUSE

Keywords

Time Series

Population

Trends

Supervisor

Mrs Pagan’a