Machine Learning in Civil Engineering - on the example of prediction of the coefficient of permeability

Abstract

This paper investigates the application of the machine learning techniques in the civil engineering, focusing on the prediction of permeability coefficient. Permeability coefficient is an important parameter in various civil engineering projects including groundwater flow analysis, soil stabilisation and geotechnical engineering. Traditional methods for estimating permeability are time-consuming and often based on laboratory tests. The machine learning offers a promising approach to predict it more efficiently and accurately. This paper studies several machine-learning techniques, verifying their applicability to predict the permeability coefficient for sands. The article analysed the predictive performance of the artificial neural network (ANN), the random forest (RF), the gradient boosting (GB) and the linear regression (LR). The most accurate algorithm in this case turned out to be the gradient boosting for which the coefficient of determination was 0.995, the mean absolute error was less than 0.001 and the root mean square error was 0.001.