Prediction of the compressive strength of concrete using various predictive modeling techniques

Concrete is one of the most essential construction materials used in construction industry. For assessment of concrete strength and quality, compressive strength is the most frequently utilized parameter. This paper establishes the application of Gaussian process, M5P model, random forest and random tree techniques for appropriate proportioning of the concrete mixes. The models proposed were based on six input parameters, namely cement, sand, coarse aggregate, water, curing period and fineness modulus, while the compressive strength was an output parameter. Five most popular statistical parameters such as Pearson correlation coefficient, mean absolute error, root mean square error, Scattering Index and Nash-Sutcliffe model efficiency were used for the assessment of the developed models. On comparison, it was found that better results were achieved with Radial bases kernel function based Gaussian process regression model as compared to other applied models. The suggested models are expected to save cost of materials, cost of labor work, time and contribute to greater accuracy. The concrete designed is anticipated to have more durability and therefore be more economical.