Qualitative Modeling of River Water Using Multivariate Adaptive Regression Spline and Evolutionary Polynomial Regression Models
Qualitative Modeling of River Water Using Multivariate Adaptive Regression Spline and Evolutionary Polynomial Regression Models
Mahdi Hosseingholizadeh1 Mohammad Najafzadeh2
1) Student of Engineering and Water Resources Management, Graduate University of Advanced Technology, Iran:
2) Associate Professor of Water Engineering Department, Graduate University of Advanced Technology,
IOI Code
:
XGKB-KEGGH
Publication :
Conservation of Natural Resources & Environment along with Forests of Iran(envprouma.ir)
Abstract :
To investigate the quality of river water, we use ecological indicators, including the macrophyte index for river. Macrophytes are aquatic plants that can be seen with the naked eye. In this research, We studied the data of 200 places in the rivers of Poland, which includes 12 water quality parameters: hydrogen potential, electrical conductivity, alkalinity, phosphate, total phosphorus, nitrate, nitrite, ammonium, organic nitrogen, total nitrogen, dissolved oxygen and biochemical oxygen demand and two hydromorphological indices including habitat quality assessment and habitat modification score as well as macrophyte index for river. We considered water quality parameters and hydromorphological indicators as independent variables and macrophyte index for river as dependent variable. Using artificial intelligence models, multivariate adaptive regression spline and evolutionary polynomial regression, we created a relationship between independent variables and dependent variable, and using the data of independent variables, we predicted the macrophyte index for river. Using statistical indices of correlation coefficient, root mean square error and mean absolute error, we investigated the performance of the models. The multivariate adaptive regression spline model with values (correlation coefficient 0.5254 and mean absolute error 9.4006) has superior performance in comparison with the evolutionary polynomial regression model with values (correlation coefficient 0.2367 and mean absolute error 16.7853). In comparison with the previous research models, the multivariate polynomial regression model with a coefficient of determination of 0.580 compared to other models has the best fit and also has a simple structure.
Keywords : multivariate adaptive regression spline evolutionary polynomial regression macrophyte index for river artificial intelligence model