FORECASTING CONCEPTUAL COSTS OF BRIDGE PROJECTS USING NON-PARAMETRIC REGRESSION ANALYSIS

Analyzing angle crashes at unsignalized intersections using machine learning techniques

Prediction of the shear modulus of municipal solid waste (MSW): An application of machine learning techniques

Developing Bicycle-Vehicle Crash-Specific Safety Performance Functions in Alabama Using Different Techniques

The improvement in blast fragmentation material size aids downstream operation efficiency by supplying a high percentage of 80% passing size of rocks (D80) material. In this study, blast fragmentation percentage prediction models were developed using five soft computing approaches, i.e., multilayer perceptron (MLP), multivariate adaptive regression spline (MARS), support vector regression SVR), decision tree (DT), and random forest (RF), combined with reptile search algorithm. Blast characteristics were obtained from 407 production blasts at the Anguran lead and zinc mine in Mahenshan, Iran. The particle size distribution resulting from blasting was determined using a digital image processing approach in the current paper. The novelty of this research lies in the integration of the reptile search algorithm (RSA) with machine-learning models for enhanced rock fragmentation prediction. In this study, the RSA was used to optimize RF model and accurate prediction of rock fragmentation due to mine blasting. The proposed models’ prediction accuracy was compared using nine performance indices such as root mean square error (RMSE), correlation coefficient (R), coefficient of determination (R 2), mean absolute percentage error (MAPE), weighted mean absolute percentage error (WMAPE), Variance Accounted For (VAF), Nash–Sutcliffe efficiency (NS), Willmott’s index of agreement (WI) and bias index (Bias). Based on the obtained results, the RF-RSA model outperformed other developed models with the lowest error and highest accuracy. Hence, the results indicated that the RF-RSA model outperforms other models in predicting blast fragmentation, achieving the lowest R 2 of 0.9734 and RMSE of 0.8386 in the training phase, and 0.9715 and 0.7464 in the testing phase. Furthermore, the model attained the lowest MAPE values of 2.3067 (training) and 2.0959 (testing), demonstrating its robustness and predictive accuracy. These findings highlight the effectiveness of the proposed AI-based approach in optimizing blast fragmentation prediction, contributing to improved operational efficiency in mining.

Surface longwave net radiation (LWNR) is a vital component in the surface radiation budget. Major progress has been made in the estimations of clear-sky LWNR. However, the estimation of cloudy-sky LWNR remains a significant challenge. In this paper, a linear model (LM) and a multivariate adaptive regression spline (MARS) model were developed to estimate the cloudy-sky LWNR from a satellite-derived surface shortwave net radiation product. Spatially and temporally matched satellite data and ground-measured LWNR, which was collected at 24 sites from four networks, were used to build and validate the linear and MARS models. The effects of land cover, climate type, and surface elevation on the estimate of LWNR were also analyzed. The MARS model, incorporating the normalized difference vegetation index (NDVI) and surface elevation (H) as the inputs, had the best performance. The determination coefficient, BIAS, and root mean square error (RMSE) were 0.51, 0.01 W/m2, and 26.10 W/m2, respectively. The developed model, when combined with freely distributed Global LAnd Surface Satellite (GLASS) products, showed promise for producing surface LWNR and all-sky surface net radiation.

Rice is a basic food that is consumed by almost half of the world's population, especially in Malaysia. Unfortunately, paddy productivity has recently decreased dramatically, which has hampered local rice supply and forced Malaysia to depend on rice from neighbouring nations. Therefore, the modelling of paddy production is important because it provides an insight that ensure a sufficient supply of a locally produced paddy. In this study, several multiple regression models were used to predict paddy production model in Malaysia from 1980 to 2022. The multiple regression models are then compared, and the best model is determined. The regression model used in this study are Multiple Linear Regression (MLR), Multivariate Adaptive Regression Spline (MARS) and Support Vector Regression (SVR). The developed models will be evaluated using RMSE, MAPE and Ljung-Box Test. These are excellent tools for gauging the precision of the fitted model, thus can be used to evaluate the model. Based on the study, the MARS model is better at modelling Malaysian paddy production from 1980 to 2022 since it has the smallest number of measurement errors. Therefore, MARS model is the best regression model to be used to model paddy production in Malaysia compared to MLR and SVR model. Since all the models have autocorrelation, a more effective approach and model can be presented to overcome the autocorrelation issue in the future.

When analyzing experiment data, the main statistical procedure is finding relationships in the data and developing models to describe them. Modeling is a powerful means of estimation, simulation, and prediction of the experiment and an indispensable tool in practical engineering. This paper considers the use of the multivariate adaptive regression splines (MARS) method to describe local relationships in the experiment data for several types of metal vapor lasers. The investigated data has been collected from patented UV ion lasers developed at the Bulgarian Academy of Sciences in the last ten years. A short description of MARS is provided. Specific MARS models have been developed describing the efficiency of an ultraviolet ion copper bromide vapor laser dependent on the change of 9 main input physical laser characteristics. The models are of linear and nonlinear type. It has been established that the best statistical indices and model fit are those of nonlinear models which contain second order local terms. The models account for about 95% of the data. The models are used to analyze the local behavior of the efficiency of various devices from the types of lasers under investigation. Graphs illustrating the established relationships are provided. The ability to predict future experiments is demonstrated.

This research deals with the subject of concrete bridges projects cost control during the design phase. This research falls within the framework of the road bridges projects where cost control is used as a tool to help design and decision making.The objective of this study is to develop a methodology to control the cost of bridge projects during the early design stages, by investing the database with any helpful software tool, the proposed cost control methodology is based on modeling the original elements and proposing a method of cost estimate based on the estimate of quantities from general parameters of origin and prices be updated periodically. The proposed methodology studies the various technical solutions chosen by the designer and evaluates them economically.In this paper, a study was conducted on (12) bridges within the capital Sana'a bridges project and (7 bridges) those carried out in Sana'a-Aden road, by using the statistical program (SPSS) as a specialized program in this field, logical and theoretical issues have also been taken into account. Results that require us to study the design of the bridges based on a controlled database and applied to the case of cost and time, and on the impact of implementation risk on the cost of the project during the design phase, and the various models necessary to design a database structure that was enriched with information collected from different bridge projects were presented in the Republic of Yemen.

PROPOSE. Development and validation of a prediction model for shared decision making for patients with lumbar spinal stenosis

One of the significant challenges during wellbore drilling is accurately predicting frictional pressure losses in symmetrical drill pipes. In this work, a Bayesian regularized neural network (BRANN) and multivariate adaptive regression splines (MARS) are employed to develop accurate and interpretable models for predicting frictional pressure losses during drilling. Utilizing data of frictional pressure loss collected through experimentation, the models are created. The model inputs include mud flow rate, mud density, pipe diameter (inside and outside diameters), and viscometer dial readings, while pressure loss is the output. Statistical comparisons between the model predictions and the actual values demonstrate the models’ ability to reasonably forecast frictional pressure losses in wells. The performance of the models, as measured by error metrics, is as follows: BRANN (0.999, 0.076, 16.76, and 11.67) and MARS (0.998, 0.0989, 21.32, and 16.499) with respect to the coefficient of determination, average absolute percentage error, root mean square error, and mean absolute error, respectively. Additionally, a parametric importance study reveals that, among the input variables, internal and external pipe diameters are the top predictors, with a relevancy factor of −0.784 for each, followed by the mud flow rate, with a relevancy factor of 0.553. The trend analysis further confirms the physical validity of the proposed models. The explicit nature of the models, together with their physical validation through trend analysis and interpretability via a sensitivity analysis, adds to the novelty of this study. The precise and robust estimations provided by the models make them valuable virtual tools for the development of drilling hydraulics simulators for frictional pressure loss estimations in the field.

Multivariate adaptive regression splines (MARS) technique is an adaptive non-parametric regression approach which has been used for various forecasting and data mining applications in recent years. This technique is more useful when a large number of explanatory variable candidates need to be considered. In this paper, the MARS technique is applied to forecast the hourly Ontario energy price (HOEP). The MARS models are developed in this work considering two scenarios for the explanatory variables. In the first scenario, the model is build based solely on the lagged values of the HOEP. In the second scenario, current and lagged values of the latest predispatch price and demand information, made available by the Ontario Independent Electricity System Operator (IESO), are also considered as explanatory variables. The forecasts generated by the developed models for high and low demand periods are significantly more accurate than the currently available forecasts for HOEP, demonstrating the MARS capability for electricity market price forecasting.

Estimating water supply through irrigation canal distribution systems is a crucial process for better water management in the irrigation schemes. This study aimed to develop an approach to predict the discharge delivered to unregulated irrigation canals (such as quaternary canals) from geometric and hydraulic information of regulated section of the system (such as tertiary and others) using machine learning approaches. The prediction performance of four Caret-based Supervised Machine Learning Models, namely; Multivariate Adaptive Regression Splines (MARS), Artificial Neural Networks (ANN), Random Forest (RF), and Radial Basis Support Vector Machines (SVM), were developed in the R programming environment, followed by variability assessment among canal outlets at Koga irrigation Scheme. Water delivery performance at quaternary canals showed a significant flow variation among the canal outlets. The comparative study of model prediction results showed identified MARS as the optimal model, both at the training stage (RMSE = 0.074 & R2 = 0.86 with normalized data) and testing stage (RMSE = 3.89 & R2 = 0.85 with rescaled data). Furthermore, the model building process and output equations of MARS were relatively interpretable compared to neuro and tree-based models, such as Artificial Neural Network and Random Forest. Thus, the MARS model was recommended to estimate the water supply to ungated irrigation canals as a function of flow rate information at gated distributary canal and other field data at lower components of irrigation schemes.KeywordsCanal outletMachine learningR environmentMARS

This study aims to build an artificial intelligence (AI)-based inference model to predict the coefficient of performance (COP) for refrigeration equipment under various R404A refrigerant conditions. The proposed model, the evolutionary multivariate adaptive regression splines (EMARS), is a hybrid of the multivariate adaptive regression splines (MARS) and the artificial bee colony (ABC). In the EMARS, the MARS primarily addresses the learning and curve fitting and the ABC carries out optimization to determine the fittest parameter settings with minimal prediction error. A tenfold cross-validation method was used to compare the performance of the EMARS against four other AI techniques, including the back-propagation neural network, classification and regression tree, genetic programming, and support vector machine. An analysis of comparison results supports EMARS as the best model for predicting the COP, with an MAPE value $$<$$<1%. In addition to performing significantly better than the four benchmark models, EMARS is unique in being able to: operate automatically without human intervention or domain knowledge; explore the approximate function of the input---output relationship; and identify the relative importance of various factors of influence automatically.

This study proposes an artificial intelligence (AI) model to predict the compressive strength and splitting tensile strength of rubberized concrete. This Evolutionary Multivariate Adaptive Regression Splines (EMARS) model is a hybrid of the Multivariate Adaptive Regression Splines (MARS) and Artificial Bee Colony (ABC) within which MARS addresses learning and curve fitting and ABC implements optimization to determine the fittest parameter settings with minimal prediction error. K-fold cross validation was utilized to compare EMARS performance against four other benchmark data mining techniques including MARS, Back-propagation Neural Network (BPNN), Radial Basis Function Neural Network (RBFNN), and Genetic Programming (GP). Comparison results showed EMARS to be the best model for predicting rubberized concrete strength and study results demonstrated EMARS as a reliable tool for civil engineers in the concrete construction industry.

Surface radiation balance plays a vital role in the earth surface system and affects many biogeophysical processes. As one of components of surface energy balance, longwave downward radiation (LWDR) is considered as the most poorly estimated radiation component, and its uncertainty is regarded as substantially larger than other terms of surface energy budget. In this paper, we applied the multivariate adaptive regression splines (MARS) method to derive LWDR based on MODIS thermal infrared bands top of atmosphere radiances and ground-based LWDR measurements. In model fitting process, the RMSE, bias and R-square value are 25.49 W/m <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> , -0.000 W/m <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> and 0.88, respectively; and in model validation stage, the RMSE, bias and R-square value are 25.63 W/m <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> , 0.481 W/m <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> and 0.87, respectively. The newly proposed model demonstrates comparable accuracy with other LWDR estimating methods and proves that MARS method is very useful in remote sensing based LWDR estimation.