Usman, Yuliana (2025) A MACHINE LEARNING APPROACH FOR PREDICTING THE COMPRESSIVE STRENGTH OF BASALT FIBER REINFORCED CONCRETE (BFRC). Undergraduate thesis, UNIVERSITAS DIPONEGORO; FAKULTAS TEKNIK.

Full text not available from this repository.

Abstract

The prediction of compressive strength in basalt fiber reinforced concrete (BFRC)
is essential for optimizing material performance and ensuring structural reliability.
This study employs a machine learning approach to develop predictive models
capable of estimating the compressive strength of BFRC based on key influencing
factors. The research explores two distinct feature selection methods including
Pearson correlation and Decision Tree Feature Importance to enhance the model’s
predictive accuracy. Two regression models, Linear Regression (LR) and Decision
Tree (DT), were developed and evaluated using multiple performance metrics,
including Root Mean Square Error (RMSE), Mean Absolute Error (MAE), Mean
Absolute Percentage Error (MAPE), and R-Squared (R2). The Decision Tree model
demonstrated superior predictive accuracy, achieving an R2 of 0.9189, RMSE of
3.0553, MAE of 2.1029, and MAPE of 4.9736 on the test dataset when optimized
using Decision Tree feature selection. Additionally, regression tree rules were
extracted to provide an interpretable mathematical framework for compressive
strength estimation. The results confirm that machine learning models can
effectively predict the compressive strength of BFRC, with feature selection
playing a vital role in improving model performance. The Decision Tree model, in
particular, outperformed Linear Regression, highlighting its capability to capture
complex nonlinear relationships in the dataset. These findings contribute to
advancing data-driven approaches in concrete mix design, facilitating more
efficient and reliable material engineering applications. Future research could
explore alternative feature selection methods like RFE or PCA, incorporate
additional machine learning models for improved prediction, and expand the dataset
to include more diverse BFRC compositions, fiber dosages, curing conditions, and
mix proportions.
Keywords: Basalt Fiber Reinforced Concrete, Machine Learning, Compressive
Strength Prediction, Feature Selection, Decision Tree

Item Type:	Thesis (Undergraduate)
Subjects:	Engineering > Civil Engineering
Divisions:	Faculty of Engineering > Department of Civil Engineering
Depositing User:	nurohmi pwk
Date Deposited:	27 Jan 2026 09:12
Last Modified:	27 Jan 2026 09:12
URI:	https://eprints2.undip.ac.id/id/eprint/44044

Actions (login required)

View Item