Assessing the spatial performance of forecasting models for parish population estimates

Abstract

Accurate population forecasts are critical for regional planning and resource allocation. In Uganda, population forecasts have historically been conducted at district levels, often resulting in coarse projections with limited applicability at finer administrative scales like parishes. This study evaluates the spatial performance of three forecasting models—Curve Fit Forecast, Exponential Smoothing Forecast, and Forest-based Forecast—for predicting parish-level population in Central Uganda. Using gridded population data from 2001 to 2020, trends were analyzed to project population changes by 2030. Results reveal that the Forest-based Forecasting model performed best, covering a majority of parishes with high accuracy. The findings highlight the importance of spatial population projections in enhancing the effectiveness of decentralized planning initiatives such as Uganda’s Parish Development Model. By leveraging high-resolution data and spatial modeling techniques, this study contributes to addressing the challenges of population growth and supporting sustainable development goals.

Keywords: Population forecasting, Spatial modeling, Curve Fit Forecast, Exponential Smoothing Forecast, Forest-based Forecast, Remote sensing.