AI life cycle – Transforming insights into action
Enterprises going through digital transformation infuse AI into their applications; a well-defined and robust methodology is required to manage the AI pipeline. Traditionally, a cross-industry standard process for data mining (CRISP-DM) methodology was used for ML projects, and it is important to understand this methodology before we explore the challenges faced with the implementation of an AI-driven application and how a more comprehensive AI life cycle will help with this process.
CRISP-DM is an open standard process model that describes the approach and the steps involved in executing data mining projects. It can be broken down into six major phases, as follows:
- Business understanding
- Data understanding
- Data preparation
- Model building
- Model evaluation
- Model deployment
The sequence of these phases is not strict and is often an iterative process. The arrows in the following diagram indicate...
