Enterprise Artificial Intelligence Transformation

Chapter 13The AI Modeling Process

What is vital is to make anything about AI explainable, fair, secure and with lineage, meaning that anyone could very simply see how any application of AI developed and why.

Ginni Rometty, chairman, president, and CEO, IBM

In this chapter, we will work through a simple example of a machine learning application to give readers a better understanding of how the model-building process works, expanding on the AI lifecycle discussed in Chapter 8. For our example use case, we will use a hypothetical telecom company that is setting up an AI model to solve a specific business problem of customer churn. We will go through a step-by-step explanation of how the modeling process works, why specific steps are essential, and how to avoid potential pitfalls. This chapter is not meant to show the reader how to become an AI modeling expert but, rather, to help managers and executives who will be supervising or interacting with their AI teams to better understand what these teams are doing and why, and what business decisions are involved in the modeling process.

Defining the Use Case and the AI Task

There are a few key questions to answer to define an AI use case well. The first is, what needs to be the output of an AI model or sequence of models – that is, what prediction must be made or what task must the model accomplish? The second is, what decision or activation within a business flow will the AI task enable so that it can be of value to the business? ...