|
SOFTWARE REVIEWS
|
|||
|
|||
|
SOFTWARE REVIEWS |
|||
|
||
|
NAG Statistical Add-Ins for Excel by Felix Grant Athough dedicated statistics software packages are widely available, the dominant workhorse for day-to-day desktop number crunching remains the spreadsheet--notably Microsoft's Excel. While such programs are not ideally suited for statistical analysis, they are familiar and trusted tools for most quality professionals and require no new learning curve. The designers of statistical packages recognize these facts of life and mimic the spreadsheet at least in appearance and basic interface. Beyond that minimum level, in fact, one way to classify such programs is the extent to which they pursue their own function-oriented agenda or bow to spreadsheet--and Excel--assumptions, both in their design and their view of data. Spreadsheets provide basic statistical functions, largely standardized from one product to another. Excel offers additional reach through the optional add-in Analysis ToolPak. The enterprising Excel user can, for instance, access tools extending to ANOVA; correlation and regression; F, t and z tests; exponential smoothing; a sampling prompt; some Fourier analysis; and component survival modeling. Nevertheless, serious and sustained statistical activity may require something stronger in the interests of efficiency. Where, then, does the quality worker go when searching for additional statistical power but not necessarily a fully dedicated package? The answer, surprisingly, comes from far away, on the other end of the scale entirely--in the world of mainframes. Leapfrogging beyond the territory of statistical software into heavier use, the programming of purpose-built solutions takes over. Fortran still dominates here, and no Excel user will want to tangle with that, but no need--the Numerical Algorithms Group offers its most widely useful statistical Fortran routines as an Excel plug-in. In the NAG Statistical Add-Ins for Excel package, these DLLs are made available to Excel through the same add-in architecture as Microsoft's own Analysis ToolPak. The collection comes on a CD, installation is pleasantly trivial, and all routines work efficiently and faultlessly. From a user viewpoint, Excel simply goes on working as always but with greatly enhanced statistical power. The 51 routines are accessed in the usual Excel way through function Wizard or by typing manually and are divided into four logical "books." The books are backed up by NAG Extract, a Visual Basic macro stored as an .xla file. Book number 0 contains fundamental material referred to in aggregate by NAG as statistical support routines. Summary measures extend to embrace skewness, kurtosis and a five-point median summary. A clutch of probability functions are here, too: beta, gamma, F ratio, normal Gaussian, Student's t and chi square, with accompanying routines to calculate their associated percentage points. Chi square statistics are also provided for two-way contingency tables. This set, on its own, makes Excel vastly more productive. The next two books together cover widely used modeling methods, with correlation and time series split from analysis of experiments and generalized linear models. Correlation and covariance matrixes are computed with optional weighting, and regression is provided by a multiple linear model. A collection of three functions cover a range of ARIMA requirements from preliminary estimation to seasonal forecasts. Block, Latin square and balanced factorial designs are analyzed in a set of ANOVA-based routines. Other options include spectral sample smoothing. The final book brings in a range of widely used multivariate tools: principal components, factors, clusters and discriminant analysis, and multidimensional scaling. NAG Extract locates as a menu item on the Excel menu bar, but can also be called from a keyboard shortcut. In use, data is extracted from an array in the worksheet (delivered by a NAG function) and relocated to a chosen destination cell or range. Various data presentation options are available, including vector transposition and triangular matrix handling. This collection will automatically be invaluable wherever Excel is used for any statistical purpose. At one end of the scale, it could obviate the need for any more specialist tools. At the other extreme, it will improve overall efficiency even where dedicated packages are in use. Either way, it is one of the very rare products whose acquisition must be good economics even before its eventual role has been evaluated. Quality managers or consultants wishing to extend their statistical reach beyond Excel but wary of abandoning their existing Excel learning investment should buy it without delay. About the author Felix Grant is a freelance journalist, lecturer and research consultant in the United Kingdom. Nag Statistical Add-Ins for Excel System requirements: Microsoft Excel 7.0 or Excel 97 running under Windows 95 or NT. Price: $149 Contact: NAG Inc. For international offices, visit Integrated SPC and Gage Tracking InfinityQS SPC Suite 6.0 by John O. Brown This software bundle from Lyle-Kearsley Systems is comprised of InfinityQS SPC, InfinityQS GTS and InfinityQS MSA. It is an excellent place to start for a package that not only can do your statistical process control but also track your gages and perform measurement systems analysis. The package is fully ODBC-compliant, with full capabilities to run from databases such as MS Access, FoxPro, Sybase and Oracle. This is a major benefit for those companies moving to an enterprise management system based on high-end software, without leaving behind those who need an integrated product with its own data set management capabilities. Each module also has the capability to run in a paperless environment. The package has full OLE capabilities and can display in one place complete information on the process being run. It can be set up to display parts drawings, process flows, control plans and instructions for processes without leaving the program. Documentation provided along with the CD-ROM tutorials and online help is very complete and guides users through the installation, use and data set modifications and alarms. The 500-page manual prompts users through the program's software functions, providing a well-rounded solution to statistical and gage control. A good set of utilities allows for data import and export, and synchronization of data with other programs. InfinityQS SPC contains a series of modules that can fit into almost any organizational function. The ability to enter data directly from gages on the shop floor is enhanced by the capabilities to enter data from keyboard, serial port or DDE. Short-run and long-run process monitoring and analysis make it a very flexible and adaptable program. The ability to real-time monitor more than 500 processes should allow for process management that would more than satisfy any data-hungry manager. The SPC module contains several nice features, such as the ability to e-mail notices of selected conditions, tell you what to adjust in your system and suggest by how much the adjustment should be made. It includes more than 300 variations of control charts. The charts and analysis include the full range of X-bar; range; standard deviations; np, p, c and u charts; and special charts such as Cp, Cpk, Ppk, Pp, moments of distribution and ANOVA, as well as many others. The programs support short-run charts popularized by the International Quality Institute and in the ASQ Quality Press book Innovative Control Charting: Practical SPC Solutions for Today's Manufacturing Environment. For a complete look at short-run processes, InfinityQS includes target charts, standardized charts and short-run charts, as well as CUSUM, defects per million opportunities and group charts. In the short-run, quick-change multisetup processes that many companies run today, this is a very powerful, necessary addition to SPC programs. The InfinityQS GTS is fully integrated with InfinityQS SPC and InfinityQS MSA. The gage tracking system can be alarmed so that if the wrong gage or an uncalibrated gage is used, the operator or others are notified of the problem. When calibrations are complete, the system issues itself a calibration certificate and then tracks due dates. Each gage can be assigned to locations, personnel or companies as needed for tracking. Security levels can be set to satisfy the document control requirements and data control requirements of ISO 9000, QS-9000 and others to make the system audit-friendly. InfinityQS MSA is as well-integrated as all the applications in this suite. It has an identical data entry system to ease its use with shop-floor studies. MSA can generate five different report types, along with the control chart analysis of the SPC package. Some of the report types are gage R&R summaries, total variation and total tolerance. The program was written to be in full compliance with the Automotive Industry Action Group's MSA publication. Tech support was good. There was a simple need for support prior to the review; we received a bad set of media. The support number was answered quickly, media was sent overnight and arrived early the next morning, and a follow-up call ensured that the new media ran OK. In summary, this suite contains all the data-management capabilities necessary for the typical manufacturing shop environment. This package is tightly integrated and gives users one interface to learn and understand. Quality professionals looking for solutions to their shop-floor control would do well to consider this software system. About the author John O. Brown is an independent quality management systems designer. He is a fellow of ASQ, and a CQA and RAB-certified quality systems lead auditor. Brown has more than 30 years' industrial experience in quality management. InfinityQS SPC Suite 6.0 System requirements: Pentium 133 or better PC; Windows 95, 98 or NT 4.0; 32 MB RAM; 15 MB free disk space; any Windows-compatible network is supported. Price: $1,795 for single user. Multiuser discounts are available. Contact: Lyle-Kearsley Systems |
|
| [QD Online] [Software] [Books] [First Word] [Applications] [SPC] |
|
|
