Statistics Software Qualification: Reference Data Sets - Hardcover

BUTLER, B; COX, M; ELLISON, S; HARDCASTLE, W

Hardcover

ISBN 10: 0854044221 ISBN 13: 9780854044221

Publisher: Royal Society of Chemistry, 1996

Synopsis

What is the best way for practising analysts to assess the fitness for purpose of their statistics processing routines? This book, published as part of the Valid Analytical Measurement initiative, provides reference data sets which will form an essential tool for the validation of the major post-analysis computational processes in analytical chemistry. Statistics Software Qualification will enable readers to test implementations of many of the important algorithms used in statistics and chemometrics. The quality of the results returned by these algorithms can vary widely. As the use of automated instruments in analysis increases, the need to detect possible deficiencies in software has become vital. This timely book provides the reader with a tool to bring a crucial part of his data analysis under the control of a quality assurance programme. The book presents a general method for generating test data, describes a step-by-step testing methodology with worked examples, and includes over 170 data sets for testing the most important statistical procedures. It also discusses the main algorithms in common use, their strengths and weaknesses, and methods of overcoming poor performance in existing software. All the data sets described in the book are also provided on disk. Statistics Software Qualification will prove an essential tool for practising analytical chemists in academia or industry, as well as instrument manufacturers developing new technology.

About this title

Review

"The authors ... have presented the material in a good user friendly manner." (Trends in Analytical Chemistry (TrAC) Vol 17 No 5 May 1998)

"... Even the non-expert user should find it straightforward to use these reference data sets to validate software." (Accreditation and Quality Assurance Vol 3 March 1998)

"...Indispensable for checking the statistical procedures on which computer software used by the analysts rely." (The Analyst, November 1997, Vol 122)

Edition details

• Publisher: Royal Society of Chemistry
• Publication date: 1996
• Language: English
• ISBN 10: 0854044221
• ISBN 13: 9780854044221
• Binding: Hardcover
• Edition number: 1
• Number of pages: 317