Scribd
Upload
6 views1 page

Machine Learning for HMA Modulus Prediction

The document discusses the limitations of two models for predicting dynamic modulus of HMA. It explores using machine learning techniques like bagging tree ensembles to improve predictions by training on data without prior knowledge of relationships. Bagging tree ensembles combine decision trees, neural networks, and other learners to improve stability and model fits.

Uploaded by

Makorogo Barugahare
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
6 views1 page

Machine Learning for HMA Modulus Prediction

The document discusses the limitations of two models for predicting dynamic modulus of HMA. It explores using machine learning techniques like bagging tree ensembles to improve predictions by training on data without prior knowledge of relationships. Bagging tree ensembles combine decision trees, neural networks, and other learners to improve stability and model fits.

Uploaded by

Makorogo Barugahare
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd

The limitations of the Witczak and Hirsch models are majorly associated with the nature of their

development, regression analysis. Obtaining highly accurate models with this method can be

difficult, especially if the equation being fit to the data is not a correct representation of the

system. To deal with the regression analysis-based limitations, a growing number of researchers

employ machine-learning techniques for statistical analysis. These techniques are capable of

“learning” the underlying behavior of a system from a set of training data without prior

knowledge of the nature of the relationships between the variables. Neural fuzzy sets, neural

networks, genetic programming and support vector machines are among the most common

machine learning techniques used in civil engineering [17, 21-24]. This study investigated the

potential of a bagging tree ensemble, another machine-learning technique for the prediction of

the dynamic modulus of HMA. Ensemble algorithms are designed to improve the stability and

model fits of base learners by aggregating their outputs to produce a single prediction. Decision

trees, neural networks, and support vector machines are among the frequently used base learners

in ensemble algorithms [25, 26]. These learners oftentimes do not perform well by themselves

either because of a high bias or excessive variance [27-29]. Ensemble algorithms have been used

in civil engineering applications like pavement crack rating, modeling concrete slump flow and

concrete dam deformation [30-32] with success. However, they have not been used for the

prediction of the dynamic modulus of HMA.

You might also like