Evaluation of Various Machine Learning Methods to Predict Istanbul’s Freshwater Consumption

Planning, organizing, and managing water resources is crucial for urban areas metropolitans. Istanbul one of the largest megacities, with a population over 15 million. The large volume demand increasing scarcity clean make long-term planning necessary this city, as sustained supply requires large-scale investment projects. Successful plans require accurate projections forecasting freshwater demand. This study considers different machine learning methods Istanbul. Using monthly consumption data provided by municipality since 2009, we compare accuracies ARIMA, Holt-Winters, Artificial Neural Networks, Recursive Long-Short Term Memory, Simple Recurrent Network models. We find that best predicted Multi-Layer Perceptron Seasonal ARIMA. From predictive modeling perspective, result another indication combined usage conventional models novel techniques to achieve highest accuracy.