machine learning defect density prediction model

Built from the world’s most comprehensive defect density prediction benchmarking study

Request more information

Since 1993, Mission Ready Software has built the world’s most extensive database of software failures and their root causes. We’ve benchmarked almost 700 factors against actual defect density on hundreds of programs, giving us unparalleled insight into what truly drives on time, on budget, functioning software. This document presents the data behind our machine-learning models, which are the engine of Requs AI Predict.

Software Project Risk Identification

The Cold Hard Truth About Software Defects 

Beyond Process: The True Drivers of Good Software

Our research reveals that highly reliable software is the result of much more than a good process. While a solid software development process is essential, it is not sufficient on its own. We’ve found that success depends on a holistic combination of five factors:

  • People and Teams: Co-locating engineers, ensuring they have a deep understanding of the product and industry, and placing them close to the target system are all instrumental to success.

  • Technical Methods: Specific techniques like fault management design and fault injection testing are crucial, yet they are not always part of a standard process.

  • Inherent Risks: Some software is inherently more difficult to develop than others. When teams face a steep learning curve, it introduces risk. Too many risks in a single release can lead to failure, even with a best-in-class organization.

  • Execution and Management: The daily practices of a team matter. Techniques like small release cycles, daily stand-ups with engineers, and detailed schedules for engineers are all execution factors that heavily influence reliability.

  • Software Process: A well-defined process is a necessary foundation, but it must be supported by the other four factors to be truly effective.  An overkilled process can steal resources needed for fault tolerant design and fault injection testing.

Uncovering Hidden Truths

Our data also challenges common assumptions. It shows that some of the most popular software development practices are not as effective at reducing defects as many people believe. By knowing which factors are ineffective, you can free up valuable time and resources to focus on the ones that truly matter. Our research proves that anything that reduces the likelihood of a project being late also reduces the number of defects that escape into the operational environment.

While everyone else has opinions, we provide the facts. This publication is the foundation of the Requs AI Predict defect density prediction software.

Cold Hard Truth About Software Defects

Edition 7a
$ 99
  • We have been benchmarking defect density since 1993 and this is our report against 679 development factors and 100+ real software programs
  • The facts about what really reduces defects - and what doesn't
  • Some of the development practices that have been popular for years actually don't reduce defects while other less popular methods have a profound effect
  • We have the facts. Everyone else has opions.
  • The study behind the machine learning defect density prediction model
Purchase here

TERMS & CONDITIONS

As per the terms and conditions page of this website, software training classes are non-refundable.

Click here for detailed terms and conditions