Machine Learning for Civil and Environmental Engineers

10 Recommendations, Suggestions, and Best Practices

What would a mistake here be like?

Ludwig Wittgenstein

Synopsis

Now that we are inching toward the end of this textbook, I find it timely to list and talk about some recommendations, suggestions, and best practices that can come in handy in a ML-based investigation. These are not meant to be followed to the teeth but rather are presented herein as I have found them to be of aid in some, if not most, of my group’s work, as well as those I had the pleasure of meeting/reading their works.¹

Before we dive into this chapter, let us start by distinguishing between what this chapter refers to as recommendations, suggestions, and best practices:²

• A recommendation is an advice that has been duly noted in the open literature. Often, a recommendation comprises a general statement.
• A suggestion is a proposition of opinion (and may largely stem from personal experiences).
• A best practice is a trend that has been commonly accepted by publications in our domain. The origin of a best practice could have been arrived at systemically, such as through a sensitivity study or empirical analysis or via heuristics (rules of thumb). Best practices tend to vary across timeframes and groups.

10.1 Recommendations

The following is a list of recommendations that could be of aid to civil and environmental engineers interested in adopting ML.

10.1.1 Continue to Learn

My favorite recommendation revolves around learning. At a fundamental level, we ...