Fig. 7From: Short-term water demand forecasting using hybrid supervised and unsupervised machine learning modelModels forecasting water outflow 8 h ahead on a weekend-day in November 2017. This figure displays 8 h forecast results of the models when implemented and tested on a held back dataset for a random weekend-day on November 2017. Analogously to the 1 h forecast in Fig. 6, Hybrid models forecasting 8 h ahead have performed better compared to SARIMA and standalone RT models. Also, Hybrid models with more neurons outperformed those with less neurons on the tested datasetBack to article page