Browse Prior Art Database

Management Data Projection Scheme

IP.com Disclosure Number: IPCOM000039441D
Original Publication Date: 1987-Jun-01
Included in the Prior Art Database: 2005-Feb-01
Document File: 1 page(s) / 12K

Publishing Venue

IBM

Related People

Gaffney, JE: AUTHOR [+2]

Abstract

A computer method is disclosed for the estimation of the density of errors which may be introduced into an element of software during its development. Histograms of the error density versus the development phase have been found to have a shape resembling a Rayleigh curve. By inspection, the area underneath a curve, representing the error density versus development phase, will give the total number of expected errors over the life of product development. In accordance with the invention, the user can take actual measurements of the occurrence of errors during several phases of the development cycle for a software product. The user can then make a best fit of a Rayleigh curve to the measured data.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 52% of the total text.

Page 1 of 1

Management Data Projection Scheme

A computer method is disclosed for the estimation of the density of errors which may be introduced into an element of software during its development. Histograms of the error density versus the development phase have been found to have a shape resembling a Rayleigh curve. By inspection, the area underneath a curve, representing the error density versus development phase, will give the total number of expected errors over the life of product development. In accordance with the invention, the user can take actual measurements of the occurrence of errors during several phases of the development cycle for a software product. The user can then make a best fit of a Rayleigh curve to the measured data. After having once found a particular Rayleigh curve which most closely fits the measured data, the user can compute the area under the curve which will represent the theoretical expected total number of errors which are estimated to occur over the entire development cycle for the software project. The use of the Rayleigh curve to represent the application of manpower during the software development process has been described in [*]. The problem solved by the invention disclosed herein is how to do curve fitting when the measured error density for a given phase of development is not the full complement of the errors which are to be expected for that phase. In accordance with the invention, the user takes the logarithm of the ordinate for each of the several differently-shaped Rayleigh curves and also takes a logarithm of the measured values for the error density from t...