Developing accurate soft computing methods for groundwater level (GWL) forecasting is essential for enhancing the planning and management of water resources. Over the past two decades, significant progress has been made in GWL prediction using machine learning (ML) models. Several review articles have been published, reporting the advances in this field up to 2018. However, the existing review articles do not cover several aspects of GWL simulations using ML, which are significant for scientists and practitioners working in hydrology and water resource management. The current review article aims to provide a clear understanding of the state-of-the-art ML models implemented for GWL modeling and the milestones achieved in this domain. The review includes all of the types of ML models employed for GWL modeling from 2008 to 2020 (138 articles) and summarizes the details of the reviewed papers, including the types of models, data span, time scale, input and output parameters, performance criteria used, and the best models identified. Furthermore, recommendations for possible future research directions to improve the accuracy of GWL prediction models and enhance the related knowledge are outlined.

This research presents a new hybridized evolutionary artificial intelligence (AI) model for modeling depth scouring under submerged weir ( d s ). The proposed model is based on the hybridization of the Extreme Gradient Boosting (XGBoost) model and genetic algorithm (GA) optimizer. The GA is hybridized to solve the hyper-parameter problem of the XGBoost model and to recognize the influential input predictors of d s . The proposed XGBoost-GA model is developed based on the incorporation of fifteen physical parameters of submerged weir. The feasibility of the XGBoost-GA model is validated against several well-established AI models introduced in the literature in addition to a hybrid XGBoost-Grid model. Several statistical performance metrics is computed for the modeling evaluation in parallel with a graphical assessment. Based on the attained prediction results, the proposed model revealed an optimistic and superior predictability performance with a maximum coefficient of determination ( R 2 = 0.933) and a minimum root mean square error ( RMSE = 0.014 m). In addition, the XGBoost-GA model demonstrated reliable feature selection for the essential physical parameters. The fifteen parameters are re-scaled to seven parameters based on their essential impacts on the d s determination.

When an algorithm or a program runs on a computer, it requires some resources. The complexity of an algorithm is the measure of the resources, for some input. These complexities are usually space and time. The subject of the empirical computational complexity has been studied in the research. This article introduces GuessCompx which is an R package that performs an empirical estimation on the time and memory complexities of an algorithm or a function, and provides a reliable, convenient and simple procedure for estimation process. It tests multiple increasing-sizes samples of the user’s data and attempts to fit one of seven complexity functions: O(N), O(Nˆ2), O(log(N)) , etc. In addition, based on the best fit procedure using leave one out-mean squared error (LOO-MSE), it predicts the full computation time and memory usage on the whole dataset. Together with this results, a plot and a significance test are returned. Complexity is assessed with regard to the user’s actual dataset through its size (and no other parameter). This article provides several examples demonstrating several cases (e.g., distance function, time series and custom function) and optimal parameters tuning.