Fuzzy Predictor of Daily Average Water Consumption Per Capita for Turkey

Creative Commons License

Akdemir H., Bayındır C.

Intelligent and Fuzzy Systems 2022, İstanbul, Turkey, 19 July 2022, no.52, pp.1-8

  • Publication Type: Conference Paper / Full Text
  • City: İstanbul
  • Country: Turkey
  • Page Numbers: pp.1-8
  • Istanbul Technical University Affiliated: Yes


The amount of daily water demanded by individuals is used as a basic parameter in the design of infrastructure systems. The purpose of this study is to examine the daily average water consumption per capita (WCPC) values used as infrastructure basic design parameters and suggested by Turkish Standards across Turkey. Accordingly, one of the aims of this study is to reveal how accurately these standards predict WCPC values, the other one is to create a model predicting better with low error rate and trend-reflecting values. WCPC belonging to 2018 for 30 Turkish cities was introduced. According to this research study, the average WCPC was found out to be 126.6 liters across the country and it ranges from 67.7 to 208 liters. The average loss percent of discharge of the cities across the country is 42% while it ranges from 23% to 71%. 10 parameters for each city that had the potential to influence water consumption are the following; average temperature, maximum temperature, average precipitation, humidity, water price, population, population density, sunshine duration, tourism intensity, and industry level. Since the problem is complex, the fuzzy logic method, which is a rule-based algorithm from the soft computing methods, was preferred and found suitable on the stage of creating the predictor model. The fuzzy model was formed with an expert perspective approach. The accuracy of the values proposed by the fuzzy model and the standards were measured and compared with R2 and RMSE parameters. The coefficient of determination of Altınbilek’s values, 2013 Provincial Bank, which are Turkish Standards, and the fuzzy model were found out as -4.77, -0.55, and 0.41, respectively. The poor estimation ability of the standards has revealed the need for the model that is able to make better estimation and measurement results proved the necessity of future examination of the predictions.