A Duplex Method for Prediction of Concrete Strength Using Dimensionality Reduction

Abstract

Concrete is a mixture of a hardened particulate material known as aggregate which is fused with water and cement. Concrete is the most used man-made material. Concrete is most commonly used on construction sites while building roads, bridges, dams, etc. Since there are different types of cement, the strength of concrete might vary. In this paper, we have used machine learning regression models like Linear Regression (LR), Lasso Regression (LaR), Ridge Regression (LR), Polynomial Regression (PR), Decision Tree (DT), K–Nearest Neighbor (KNN), Random Forest (RF), Gradient Boosting Regression (GBR), ADA Boosting (ADA), Support Vector Machine (SVM) and XG Boost (XGB) to predict the strength of concrete. This paper emphasizes on dual approach for predicting the concrete strength where the first approach is based on consideration of all the features for training and in the second approach, dimensionality reduction is performed using Principal Component Analysis (PCA) technique. In terms of results, XG Boost Regression (XGB) model with all features gave the best R <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> score of 0.9206, Mean Squared Error of 20.6790, Root Mean Squared Error of 4.5474 and Mean Absolute Error of 2.8966.