Machine Learning for Civil and Environmental Engineers

4 Machine Learning Algorithms

Where others saw chaos, I saw a pattern.

Hari Seldon (Foundation, 2021)

Synopsis

This chapter presents the tools we will use to create, develop, and/or explore ML models and solutions. These tools are referred to as algorithms. There are numerous ML algorithms as well as algorithmic families.¹ Much of them are appropriate to engineering problems. Most² of which are applicable to civil and environmental engineering. Many are easy to use, and some may require a bit of work.³ I will primarily focus on algorithms that utilize tabular data and then showcase the use of deep learning and computer vision techniques. I will also show visualization tools that you can try on your own to better visualize how the presented algorithms handle and process data (i.e., features) and then arrive at their predictions. Finally, I will be attaching several code libraries⁴ and already-built scripts for each discussed algorithm, as well as in the Appendix, for you to try and apply.

4.1 An Overview of Algorithms⁵

This section provides an overview of some of the current and commonly used ML algorithms that have been successful in solving or overcoming problems and case studies belonging to the domain of civil and environmental engineering (see Figure 4.1). For simplicity, I will be presenting such algorithms in alphabetical order. The majority of the algorithms that fall under supervised learning can accommodate regression and classification problems. On the other hand, ...