PSI - Issue 70

Available online at www.sciencedirect.com

ScienceDirect

Procedia Structural Integrity 70 (2025) 440–446

Structural Integrity and Interactions of Materials in Civil Engineering Structures (SIIMCES-2025) Machine Learning Based Compressive Strength Prediction Model for Lime Mortar Using Supervised Learning Algorithms Santhosh Kumar N V a , Surendar M a, * , Sindhu V b , Ragavan V c , Lavanya Prabha S a

a Department of Civil Engineering, Easwari Engineering College, Ramapuram, Chennai, Tamilnadu, India b Research Scholar, Indian Institute of Information Technology, Tiruchirappalli, Tamilnadu, India c ATCI, Accenture, Chennai, Tamilnadu, India

Keywords: Lime mortar;Machine learning;Compressive strength prediction;Feature importance analysis;Supervised learning algorithms; Material science Abstract This study investigates the application of machine learning models for predicting the compressive strength of lime mortar with six features as inputs: Lime, Ground Granulated Blast-Furnace Slag (GGBS), Red Mud, Water, Fine Aggregate, and Mortar Age. The models were trained and tested on a dataset of 381 data points from large-scale laboratory experiments. The accuracy of the prediciton was tested with five supervised learning models: Support Vector Machine (SVM), Multiple Linear Regression (MLR), K-Nearest Neighbors (KNN), XGBoost, and Random Forest (RF). For each model, unprocessed and pre-processed (normalized) data were used as inputs, and MSE, MAE, and Adj R² as metrics for evaluation. Results indicated that MLR performed steadily in both normalized and raw data, whereas SVM performed much better when data were normalized, achieving the highest Adj R². MLR, RF, and XGBoost performed steadily independent of data scaling; however, SVM and KNN were helped by normalization. These findings have significant implications concerning machine learning's role in material science. These methods provide a shortened, data-based approach to minimize the amount of necessary time and money spent when performing experiments in construction. The paper is significant background for the structural engineering industry. © 2025 The Authors. Published by ELSEVIER B.V. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0) Peer-review under the responsibility of International Conference on Structural Integrity Organizers

* Corresponding author. Tel.: "+91 99400 15288" E-mail address: surendar.m@eec.srmrmp.edu.in

2452-3216 © 2025 The Authors. Published by ELSEVIER B.V. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0) Peer-review under the responsibility of International Conference on Structural Integrity Organizers 10.1016/j.prostr.2025.07.075