concept ML workflow in category machine learning

This is an excerpt from Manning's book Real-World Machine Learning. Login to get full access to this book.

1.3. Following the ML workflow: from data to deployment

In this section, we introduce the main workflow for integrating machine-learning models into your applications or data pipelines. The ML workflow has five main components: data preparation, model building, evaluation, optimization, and predictions on new data. The application of these steps has an inherent order, but most real-world machine-learning applications require revisiting each step multiple times in an iterative process. These five components are detailed in chapters 2 through 4, but we outline them in this introduction to whet your appetite for getting started. Figure 1.7 outlines this workflow, and the following sections introduce these concepts from top to bottom. You’ll see this figure a lot throughout the book as we introduce the various components of the ML workflow.

Figure 4.1. Evaluation and optimization in the ML workflow