African Atmospheric Sciences (Earth Science focus) | 12 April 2002
Time-Series Forecasting Model Evaluation for Measuring Adoption Rates in Municipal Water Systems in Kenya
M, w, i, h, a, k, i, K, i, n, y, a, n, j, u, i, ,, O, d, h, i, a, m, b, o, O, t, i, e, n, o
Abstract
This study examines the adoption rates of municipal water systems in Kenya by utilising time-series forecasting models to analyse trends and patterns over a decade. A suite of time-series analysis techniques was applied including ARIMA (AutoRegressive Integrated Moving Average) to analyse the data from to . Robust standard errors were used to account for forecast uncertainties. The model identified a significant upward trend in adoption rates, with an estimated increase of 8% per annum over the study period (the direction and proportion are based on the empirical data analysed). The refined time-series forecasting models demonstrated improved accuracy in predicting municipal water system adoption rates compared to previous methods. Future research should focus on integrating additional socio-economic factors into the model for more comprehensive forecasts, particularly regarding rural areas where coverage is limited. The empirical specification follows $Y=\beta_0+\beta^\top X+\varepsilon$, and inference is reported with uncertainty-aware statistical criteria.