Electrical conductivity (EC), one of the most widely used indices for water quality assessment, has been applied to predict the salinity of the Babol-Rood River, the greatest source of irrigation water in northern Iran. This study uses two individual—M5 Prime (M5P) and random forest (RF)—and eight novel hybrid algorithms—bagging-M5P, bagging-RF, random subspace (RS)-M5P, RS-RF, random committee (RC)-M5P, RC-RF, additive regression (AR)-M5P, and AR-RF—to predict EC. Thirty-six years of observations collected by the Mazandaran Regional Water Authority were randomly divided into two sets: 70% from the period 1980 to 2008 was used as model-training data and 30% from 2009 to 2016 was used as testing data to validate the models. Several water quality variables—pH, HCO₃⁻, Cl⁻, SO₄²⁻, Na⁺, Mg²⁺, Ca²⁺, river discharge (Q), and total dissolved solids (TDS)—were modeling inputs. Using EC and the correlation coefficients (CC) of the water quality variables, a set of nine input combinations were established. TDS, the most effective input variable, had the highest EC-CC (r = 0.91), and it was also determined to be the most important input variable among the input combinations. All models were trained and each model’s prediction power was evaluated with the testing data. Several quantitative criteria and visual comparisons were used to evaluate modeling capabilities. Results indicate that, in most cases, hybrid algorithms enhance individual algorithms’ predictive powers. The AR algorithm enhanced both M5P and RF predictions better than bagging, RS, and RC. M5P performed better than RF. Further, AR-M5P outperformed all other algorithms (R² = 0.995, RMSE = 8.90 μs/cm, MAE = 6.20 μs/cm, NSE = 0.994 and PBIAS = −0.042). The hybridization of machine learning methods has significantly improved model performance to capture maximum salinity values, which is essential in water resource management.

In this research, three different evolutionary algorithms (EAs), namely, particle swarm optimization (PSO), genetic algorithm (GA) and differential evolution (DE), are integrated with the adaptive neuro-fuzzy inference system (ANFIS) model. The developed hybrid models are proposed to forecast rainfall time series. The capability of the proposed evolutionary hybrid ANFIS was compared with the conventional ANFIS in forecasting monthly rainfall for the Pahang watershed, Malaysia. To select the optimal model, sixteen different combinations of six different lag attributes taking into account the effect of monthly, seasonal, and annual history were considered. The performances of the forecasting models were assessed using various forecasting skill indicators. Moreover, an uncertainty analysis of the developed forecasting models was performed to evaluate the ability of the hybrid ANFIS models. The bound width of 95% confidence interval (d-factor) and the percentage of observed samples which was enveloped by 95% forecasted uncertainties (95PPU) were used for this purpose. The results indicated that all the hybrid ANFIS models performed better than the conventional ANFIS and for all input combinations. The obtained results showed that the models with best input combinations had the (95PPU and d-factor) values of (91.67 and 1.41), (91.03 and 1.41), (89.74 and 1.42), and (88.46 and 1.43) for ANFIS-PSO, ANFIS-GA, ANFIS-DE, and the conventional ANFIS, respectively. Based on the 95PPU and d-factor, it is concluded that all hybrid ANFIS models have an acceptable degree of uncertainty in forecasting monthly rainfall. The results of this study proved that the hybrid ANFIS with an evolutionary algorithm is a reliable modeling technique for forecasting monthly rainfall.