We collect 1 030 groups of concrete compressive strength test data,and obtain a model that can be used to predict the values of concrete compressive strength by training the AdaBoost algorithm.The results show that the AdaBoost algorithm model can accurately and effectively predict the concrete compressive strength under the condition of given input variables;the average values of coefficients of determination using a 10-fold cross-validation reach 0.952,and the mean absolute percentage errors reach 11.39％,indicating that the prediction errors are very low;compared with the independent learning algorithms of artificial neural network and support vector machine,the AdaBoost algorithm shows the superiority of ensemble learning algorithms.The number of training data sets and the type of weak learners in the AdaBoost algorithm model were discussed.According to the number of input variables,it is found that using 80％of the 1 030 sets of data gives good prediction results.
关键词
混凝土/抗压强度/机器学习/交叉验证/训练数据集/弱学习器
Key words
concrete/compressive strength/machine learning/cross-validation/training data set/weak learner
万方数据