Prediction of Concrete Compressive Strength and Slump by Machine Learning Methods

Table 3

Hyperparameters of machine learning regression models.

Model	Method	Required package	Tuning parameter
Classification and regression trees (CART)	rpart	CRAN	method = “anova”
Random forest (RF)	rf	Caret	ntree = 100
Support vector machine (SVM)	svmLinear, svmPoly	Caret	gamma = 0.001, cost = 100
Partial least squares (PLS)	pls	Caret	tuneLength = 20
Artificial neural network (ANN)	mlp	RSNNS	size = 5, maxit = 100, learnFuncParams = 0.1
Bootstrap aggregation (bagging)	bagging	ipred	na.action = na.rpart; method.type = “WM”, num.labels = 7
Fuzzy logic (FL)	frbs.learn	frbs	max.iter = 30; step.size = 0.01,; gradient descent = 00.1; type.implication.func = “ZADEH”