Water scarcity and increasing water demand especially for residential end-use are major challenges facing Tunisia. The need to accurately forecast water consumption is useful for the planning and management of this natural resource. In the current study, quarterly time series of household water consumption in Tunisia is forecasted using a comparative analysis between traditional Box-Jenkins method and artificial neural networks approach. In particular, an attempt is made to test the effectiveness of data preprocessing, such as detrending and deseasonalization, on the accuracy of neural networks forecasting. Results indicate that traditional Box-Jenkins me thod outperforms neural networks estimated on raw, detrended, or deseasonalized data in terms of forecasting accuracy. However forecasts provided by neural network model estimated on combined detrended and deseasonalized data are significantly more accurate and much closer to the actual data. This model is therefore selected to forecast future household water consumption in Tunisia. Projection results suggest that by 2025, water demand for residential end-use will represent around 18% of the total water demand of the country.
- Artificial neural networks,
- time series forecasting,
- residential water demand
Available at: http://works.bepress.com/maamar_sebri/4/