Building custom ML models with Cloud AI Platform and BigQuery ML
Before getting to the different components and capabilities within Cloud AI Platform (Unified) – hereafter referred to as AI Platform – it's useful to understand the ML workflow stages. A typical production-ready workflow for a supervised ML application is represented in the following diagram:
Figure 11.5 – The ML workflow
Let's break this down.
The first and most fundamental step is that of identifying the business goals and formulating the business case. Ensure the problem is well defined by answering the following two questions at a minimum:
- What information are you trying to get out of the model?
- Why will this information be useful?
Sometimes these questions generate hypotheses to be tested. Therefore, you must determine the project's feasibility and whether you can afford to fall short of achieving the outcomes. A machine learning...
