Using an Azure Machine Learning Approach for Flexible Pavement Maintenance

Machine learning (ML) is one of the intelligent methodologies that has shown promising results in the domains of classification and prediction. In this paper, Azure Machine Learning (AML) systems were applied for flexible pavement maintenance classification problems and solutions. For prediction, four parameters were used as the inputs namely severity, density, road functional, and average daily traffic (ADT) while the output parameters were treatment techniques. This paper provides a critical analysis of classification algorithms: two-class support vector machine, multi-class decision forest, and multi-class neural network. Characteristic significance analysis was carried out to investigate how each classifier utilized the information available in the dataset, focusing on the application of azure machine learning classification (AMLC) to the pavement maintenances results prediction. Overall, the findings showed that predictions using Multi-Class Neural Network (MCNN) were obtaining an accuracy of 0.99. The results indicated that prediction obtained from AML training seemed to be more powerful, with a smaller standard error. Azure ML technique showed high accuracy with satisfactory results and capability of predicting pavement maintenance rigorously.