Machine Learning, 2nd Edition

Chapter 12Communication with Clients

Many true statements are too long to fit on a [PowerPoint] slide, but this does not mean we should abbreviate the truth to make the words fit.

— Edward R. Tufte, The Cognitive Style of PowerPoint, 2003

I have not the pleasure of understanding you. Of what are you talking?

— Mr. Bennet in Jane Austen, Pride and Prejudice, 1813

In order to produce a real-world effect, a machine learning practitioner typically works with clients or collaborators in one or more of the natural, social, or computer sciences. It is therefore beneficial to the machine learning practitioner to be able to communicate with people working in these fields. Indeed, it is the author’s view that it is more necessary for the machine learning practitioner to speak the language of their scientific clients than vice versa – we must meet them more than halfway.

The need for communication is highlighted by the flow chart shown in Figure 12.1, which illustrates the process of developing a predictive model for a client in an idealized setting.¹ Four of nine steps in the “main loop” of the flow chart (and one of three steps in the “initial loop” at the start) involve client communication. Implicit in the flow chart is the assumption that the practitioner appropriately documents their work at each step.

Figure 12.1 A flow chart showing an idealized process of developing a predictive ...