Statistics Software: examples and applications

The computer tools available to engineers have enabled a considerable expansion in statistical studies. Sample sizes, the number of variables observed and the precision of observations are no longer an obstacle to the use of complex statistical methods. Software dedicated to statistics has been developed over the last few decades, while more general-purpose software has been enriched with statistical functions. The aim of this article is to highlight their basic functionalities through a series of examples.

All users are confronted with a commercial logic: for private individuals, the budget is often limiting, whereas for corporate executives, the choices made by the company/organization are often prescriptive. With this in mind, the software products used to illustrate the practical examples have been selected on the basis of their membership of one of the various families within this prism, without prejudice to any pre-eminence over other members of the family:

• Excel: available in all MS-Office suites (under Windows or MacOS), it has paved the way for various add-ins, such as XLSTAT, marketed separately, which offer libraries of Excel "macros" dedicated to statistics: these add-ins remain dependent on the strategies of the Excel publisher;

• SAS: a comprehensive, pioneering software package, it is the benchmark for many statisticians, even if they use other programs;

• JMP IN: the "consumer" version of SAS, it competed head-on with excellent products designed for PCs while SAS reigned over the major systems (Statistica, StatLab, StatView, SPSS, Systat...);

• the "Open source" family, represented by R software, associated with free use of software (GNU license) and collegial evolution;

• Finally, this article does not take into account "local" developments, carried out by organizations or individuals. These include the free M-CARE calibration results modeling software developed by the Collège français de métrologie.

The presentation of the functionalities of these different software packages focuses on :

• basic functions ;

• fit and comparison tests ;

• single- and double-entry analysis of variance ;

• regression analysis;

• principal component analysis.