UKaRsT, Год журнала: 2024, Номер 8(1), С. 28 - 41
Опубликована: Апрель 30, 2024
Accurately predicting concrete compressive strength is fundamental for optimizing mix designs, ensuring structural integrity, and advancing sustainable construction practices. Increased demands safer, more durable infrastructure necessitate effective predictive models. This research aims to compare the effectiveness of six machine learning models such as Linear Regression, Random Forest, Support Vector Regression (SVR), K-Nearest Neighbors (KNN), Gradient Boosting, XGBoost predict strength. Used a dataset 1030 instances with varying mixture compositions, conducted extensive exploratory data analysis, applied feature engineering scaling enhance model performance. Assessments were performed 5-fold cross-validation approach R-squared (R²) metric. In addition, SHAP value used understand influence each on results. The results revealed that significantly outperformed other models, achieving an average R² 0.9178 standard deviation 0.0296. Notably, Forest Boosting also demonstrated robust capabilities. Based our experiment, these effectively predicted strengths close actual measured values, confirming their practical applicability in civil engineering. values provided insights into significant impact age cement quantity outputs. These highlight advanced ensemble methods' prediction underscore importance enhancing accuracy.
Язык: Английский