Titlebook: Why AI/Data Science Projects Fail; How to Avoid Project Joyce Weiner Book 2021 Springer Nature Switzerland AG 2021

狼群

978-3-031-00557-2Springer Nature Switzerland AG 2021

裂口

Why AI/Data Science Projects Fail978-3-031-01685-1Series ISSN 2766-8975 Series E-ISSN 2766-8967

TSH582

Synthesis Lectures on Computation and Analyticshttp://image.papertrans.cn/w/image/1028106.jpg

Enteropathic

Introduction and Background,jects failed. Looking into it, I learned that in this case, failure was defined as “not being deployed.” So, starting a project that didn’t make it to production was the definition of the project failing. I was surprised by the number. 85% is a large percentage.

吝嗇性

,,get resources and funding. It helps to answer management’s question of “what do I get?” Knowing the expected deliverables for a project helps in getting support. It also helps with defining when you are done with a project.

breadth

外露

Project Phases and Common Project Pitfalls,Let’s take a more in-depth look into the reasons for projects not to get to production. While there are many reasons, and clearly this must be true because 87% of projects don’t make it, I’ll cover 5 reasons that are systematic in nature. These reasons are:

思想

準則

Model-Building Phase,Two of the project pitfalls relate directly to the model building phase of a data science project: couldn’t explain the model, and the model was too complex. The tools to address these pitfalls are to keep things simple and leverage explainability.

Libido

Summary of the Five Methods to Avoid Common Pitfalls,The current statistic is that 87% of AI/big data projects fail. In this context, failing means that the project never reaches deployment. By applying 5 methods to avoid common pitfalls, you can give your project a better opportunity to beat the odds and be one of the 13% that make it to production. The five pitfalls are: