< back to main site

Publications

Testing functions for linear regression.

Cox, M G; Dainton, M P; Harris, P M (2000) Testing functions for linear regression. NPL Report. CMSC 08/00

[img] Text
cmsc08.pdf

Download (159kB)

Abstract

The aim of the work reported here is to contribute to the development of an infrastructure, comprising supporting information and guidelines, to ensure that the use of software, particularly spreadsheets and proprietary software packages, within metrology is made as effective as possible. This is to be achieved by reporting the results of the objective testing of the intrinsic and in-built functions included within spreadsheets and other proprietary software packages that are popular in metrology applications.
We describe the application of a general methodology for testing the numerical accuracy of software to functions for linear regression (applied to the straight-line linear regression problem) taken from a number of spreadsheet, statistical and scientific software packages, including Microsoft Excel, MathCAD, S-PLUS, Matlab, and the NAG and IMSL Fortran libraries. Each stage of the methodology, from the provision of a specification for the function tested through the definition of performance parameters and measures to the presentation and interpretation of the test results, is presented. In this way, and by stating any assumptions made in the application of the methodology,the testing undertaken is made as objective as possible.
This report constitutes one of the deliverables of Project 2.1 “Testing Spreadsheets and Other Packages Used in Metrology” within the UK Department of Industry’s National Measurement System Software Support for Metrology Programme 1998-2001.

Item Type: Report/Guide (NPL Report)
NPL Report No.: CMSC 08/00
Subjects: Mathematics and Scientific Computing
Mathematics and Scientific Computing > Numerical Computation
Last Modified: 02 Feb 2018 13:17
URI: http://eprintspublications.npl.co.uk/id/eprint/1792

Actions (login required)

View Item View Item