PSI - Issue 70

Santhosh Kumar N V et al. / Procedia Structural Integrity 70 (2025) 440–446

442

Forest, SVM, XGBoost, KNN, and MLR can all be combined to provide a complete assessment of the compressive strength of lime mortar. 2.1 Random Forest (RF) Random Forest is an ensemble learning approach where many decision trees are constructed during training and the average prediction of each of the individual trees is output for regression. Random Forest is not subject to overfitting and can be exceptionally good in high dimensional data. Support Vector Machines (SVM) is a supervised learning algorithm which can be used for classification and regression tasks but is more commonly used for classification tasks. In regression (Support Vector Regression, SVR), the aim is to investigate a hyperplane that best fits the data. SVM performs regression by attempting to minimize the prediction error. 2.3 XGBoost XGBoost (Extreme Gradient Boosting) is a highly optimized and advanced version of gradient boosting machines in terms of speed and performance. In XGBoost, an ensemble of decision trees are created sequentially and each tree corrects the previous tree's errors. 2.4 K-nearest neighbors (Knn) K-nearest neighbors (KNN) is an instance-based supervised learning algorithm that makes predictions for a target variable by averaging the values of the k nearest neighbors in the feature space. 2.5 Multiple linear regression (Mlr) Multiple linear regression is a statistical method which investigates the relationship between multiple independent variables and a dependent variable by creating a linear equation that best fits the observed data. 3. Problem description Conventional tests prepare mortar specimens with different proportions, cure under controlled conditions, and conduct compressive strength tests at various ages. While the most precise method, it would be impossible to achieve quick material optimizations or mass applications (Zhang et al., 2020; Singh et al., 2021). Traditional experiments prepare mortar samples with varying compositions, cure under standardized conditions, and perform compressive strength tests at different ages. Although the most accurate method, being able to perform rapid material optimizations or large-scale applications would not be attainable (Zhang et al., 2020; Singh et al., 2021). 3.1 Descriptive statistics of input and output variables for machine learning To respond to these matters, this research applies machine learning (ML) methods to forecast the compressive strength of lime mortar, based on its composition and the age of the curing. A total of 381 data points were developed in the course of the laboratory work which concurrently collected/loading information and critical input variables/attributes such as the amount of lime, red mud, GGBS (Ground Granulated Blast Furnace Slag), water, fine aggregate and age during curing. The dependent variable, compressive strength, was measured using regular testing protocols and procedures. The dataset is diverse in atomic-level compositions and time of curing conditions, it is adequate to train and validate the ML models (Akinosho et al., 2021, Feng et al. 2022). 2.2 Support vector machine (Svm)