This study attempts to develop a machine learning model to predict the concrete slump as a function of mix proportions, taking advantage of the 3599 observations of industrially produced ready-mix concrete applied in various construction projects. Following statistical analysis and data visualization to obtain insights from the data, seven machine learning models, covering linear, non-linear, and ensemble learning techniques, are explored to predictively classify the slump into one of the eight characteristic classes. Extreme gradient boosting and random forest are found to be the two best ones after a comprehensive comparison of the seven models’ performance against metrics of accuracy, Kappa, Matthews correlation coefficient, logLoss, receiver operating characteristic plot, precision-recall plot, and the area under the curve corresponding to the two plots. The multi-classification slump estimation using industrial concrete data offers a practical and straightforward means for the industry to enhance quality control and performance assessment.

本研究试图开发一个机器学习模型来预测混凝土坍落度作为混合比例的函数，利用对应用于各种建筑项目的工业生产的预拌混凝土的 3599 次观察。在统计分析和数据可视化以从数据中获得洞察力之后，探索了七种机器学习模型，包括线性、非线性和集成学习技术，以预测性地将衰退分类为八个特征类别之一。在将七个模型的性能与准确度、Kappa、马修斯相关系数、logLoss、接收器操作特征图、精确召回图、以及对应于两个图的曲线下面积。使用工业混凝土数据进行多分类坍落度估计，为行业加强质量控制和性能评估提供了一种实用且直接的方法。