Book Image

IBM Cloud Pak for Data

By : Hemanth Manda, Sriram Srinivasan, Deepak Rangarao
3 (1)
Book Image

IBM Cloud Pak for Data

3 (1)
By: Hemanth Manda, Sriram Srinivasan, Deepak Rangarao

Overview of this book

Cloud Pak for Data is IBM's modern data and AI platform that includes strategic offerings from its data and AI portfolio delivered in a cloud-native fashion with the flexibility of deployment on any cloud. The platform offers a unique approach to addressing modern challenges with an integrated mix of proprietary, open-source, and third-party services. You'll begin by getting to grips with key concepts in modern data management and artificial intelligence (AI), reviewing real-life use cases, and developing an appreciation of the AI Ladder principle. Once you've gotten to grips with the basics, you will explore how Cloud Pak for Data helps in the elegant implementation of the AI Ladder practice to collect, organize, analyze, and infuse data and trustworthy AI across your business. As you advance, you'll discover the capabilities of the platform and extension services, including how they are packaged and priced. With the help of examples present throughout the book, you will gain a deep understanding of the platform, from its rich capabilities and technical architecture to its ecosystem and key go-to-market aspects. By the end of this IBM book, you'll be able to apply IBM Cloud Pak for Data's prescriptive practices and leverage its capabilities to build a trusted data foundation and accelerate AI adoption in your enterprise.
Table of Contents (17 chapters)
1
Section 1: The Basics
4
Section 2: Product Capabilities
11
Section 3: Technical Details

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...