Property Prediction

Machine learning has revolutionized the field of materials science by enabling accurate and efficient prediction of material properties. This chapter presents a comprehensive overview of the key steps involved in developing machine learningMachine learning modelsModels for structured data in materials science, along with an introduction to physics-informed machine learningMachine learning. This chapter highlights the challenges associated with predicting material properties and emphasizes the importance of robust computational methods. It discusses the crucial stages of constructing machine learningMachine learning modelsModels, including data preprocessing, feature engineering and selection, and modelModels training and evaluation. Furthermore, the chapter introduces how domain-specific knowledge and fundamental physical principles can be infused with machine learningMachine learning modelsModels for property prediction, an approach known as physics-informed machine learningMachine learning. This integration enhances prediction accuracy and ensures adherence to underlying material behavior laws and principles. Altogether, this chapter describes one of the most commonly used application of ML, that is, to predict the properties of materials as a function of composition based on structured data. Additional approaches towards property prediction based on microstructure images, and other semi-structured or unstructured data are discussed in detail in the following chapters.