Hard facts

The "Naked Truth About Software Engineering - Version 5", 2006, 110 pages.

Table of Contents

This document contains the latest results of our predictive modeling research and development.  It includes:

Hundreds of characteristics that have been correlated to actual fielded defects in 90 real companies developing real software for real systems.

• The characteristics that correlated the strongest to fielded defects.
• The characteristics that had the most impact in terms of magnitude on fielded software defects.
• How each of the characteristics correlated to the other characteristics.
• The relative cost to implement each characteristic.
• The relative calendar time to implement each characteristic.
• The model that evolved for predicting software defects and probability of being late.
• Average defect density by SEI CMMi
• Average defect density by application type, industry type and risk type
• Average defect density by year of development completion

The 1%, 10%, 25%, 50%, 75%, 90% and 99% groups that resulted from the modeling effort.  The practices employed by each percentile group, their average fielded defects and how much each percentile group spent relatively on software engineering.

The actual results of the 62 companies for each of the survey questions.

ADVANCED STATISTICS

The advanced statistics for the above study can be found in a separate document.   These advanced statistics are only for those with backgrounds in statistics or data mining and include:

• Intercorrelations between each of the 600 practices/characteristics
• Multi-variate statistics used to develop the Shortcut and Fullscale prediction models

Pricing