An Approach Based on the Use of Commercial Codes and Engineering Judgement for the Battle of Water Demand Forecasting

This paper demonstrates the synergistic use of engineering judgment and statistical/deep learning models, implemented through a four-step process using the software SAS Viya 4. Initial data filtering, input variable determination, and simultaneous application of RNN, LSTM, and GRU forecasting algorithms are conducted. Results are evaluated based on Battle of Water Demand Forecasting criteria, refining parameters iteratively for enhanced prediction accuracy. The methodology iteratively incorporates new data, streamlining neural network resolution.