Estimating the concrete compressive strength using hard clustering and fuzzy clustering based regression techniques

Area of Science:

• Civil Engineering
• Materials Science
• Data Science

Background:

• Accurate prediction of concrete compressive strength is crucial for construction quality assurance and material design.
• Regression techniques are standard for predicting concrete strength, but their accuracy can be limited.
• Integrating clustering with regression offers a promising approach to improve prediction accuracy.

Purpose of the Study:

• To propose and evaluate a novel cluster regression technique for predicting concrete compressive strength.
• To demonstrate that combining clustering and regression methods minimizes prediction errors.
• To compare the performance of fuzzy C-means and K-means clustering algorithms within the regression framework.

Main Methods:

• A two-stage approach was developed: first, clustering similar concrete data characteristics, and second, applying regression analysis within each cluster.
• The cluster regression technique was implemented for concrete compressive strength estimation.
• Experiments were conducted to validate the proposed method and compare clustering algorithms.

Main Results:

• The cluster regression technique demonstrated reduced prediction errors compared to traditional regression methods.
• Fuzzy clustering algorithm C-means yielded better results than the K-means algorithm in predicting concrete compressive strength.
• The proposed method effectively improves the accuracy of concrete strength prediction.

Conclusions:

• Clustering combined with regression significantly enhances the accuracy of concrete compressive strength prediction.
• Fuzzy C-means is a more effective clustering algorithm than K-means for this specific application.
• The developed cluster regression technique offers a valuable advancement for concrete material science and construction engineering.