Scholarlink

<p>The urgent need to find sustainable alternatives to Portland cement, which emits substantial CO<sub>2</sub> during production, has driven research toward exploring environmentally friendly materials. Ground granulated blast furnace slag (GGBFS) and fly ash geopolymer concrete (FA-GPC) have emerged as promising alternatives due to their potential to reduce CO<sub>2</sub> emissions and address industrial waste disposal issues. This research paper focuses on enhancing the accuracy of predicting the compressive strength of GPC containing FA and GGBFS through efficient machine learning models: Random Forest (RF) and Decision Tree (DT). The study conducts comprehensive laboratory investigations, gathering numerous samples, and using statistical measurements such as RMSE, MAE, and R-value to evaluate the models' performance. Sensitivity analysis is performed to identify key factors influencing geopolymer concrete strength, enabling informed decisions for concrete mix design optimization. The results indicate that the decision tree model generally outperforms the random forest model in accuracy and explaining data variance, particularly with smaller datasets. However, the choice between models should consider specific requirements and dataset sizes of the problem at hand for real-world applications in the construction industry. The evaluation metrics for two regression models, RF and DT, trained on different ratios of training and testing data (70:30 and 50:50), were compared. For the 70:30 ratio, the DT model outperformed the RF model with lower error metrics (RMSE, MSE, MAE) and a higher R-squared value (R<sup>2</sup>). Similarly, for the 50:50 ratio, the DT model showed better performance than the RF model. The Decision Tree models demonstrated particularly exceptional performance on the testing data, especially with D1 achieving a perfect fit with an R2 value of 1.0.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Prediction and comparison of compressive strength of geopolymer concrete by using decision tree and random forest regression model

  • 期刊论文

  • 发表日期：

    2025年8月18日

查看全文
  • Manvendra Verma,
  • Ujjwal Sharma

摘要

The urgent need to find sustainable alternatives to Portland cement, which emits substantial CO2 during production, has driven research toward exploring environmentally friendly materials. Ground granulated blast furnace slag (GGBFS) and fly ash geopolymer concrete (FA-GPC) have emerged as promising alternatives due to their potential to reduce CO2 emissions and address industrial waste disposal issues. This research paper focuses on enhancing the accuracy of predicting the compressive strength of GPC containing FA and GGBFS through efficient machine learning models: Random Forest (RF) and Decision Tree (DT). The study conducts comprehensive laboratory investigations, gathering numerous samples, and using statistical measurements such as RMSE, MAE, and R-value to evaluate the models' performance. Sensitivity analysis is performed to identify key factors influencing geopolymer concrete strength, enabling informed decisions for concrete mix design optimization. The results indicate that the decision tree model generally outperforms the random forest model in accuracy and explaining data variance, particularly with smaller datasets. However, the choice between models should consider specific requirements and dataset sizes of the problem at hand for real-world applications in the construction industry. The evaluation metrics for two regression models, RF and DT, trained on different ratios of training and testing data (70:30 and 50:50), were compared. For the 70:30 ratio, the DT model outperformed the RF model with lower error metrics (RMSE, MSE, MAE) and a higher R-squared value (R2). Similarly, for the 50:50 ratio, the DT model showed better performance than the RF model. The Decision Tree models demonstrated particularly exceptional performance on the testing data, especially with D1 achieving a perfect fit with an R2 value of 1.0.