Statistica 8 Enterprise Edition
As top of the market analytic systems get bigger and more complex, the market for their application growing and differentiating, there is a growing move to package their power in segmented ways. Of the leaders in this, StatSoft’s Statistica is prominent. The base edition of Statistica 8, carrying the core analytic tools, was released in the spring of 2007 (reviewed here in May). A major maintenance release later in the year added a range of enterprise and SPC (statistical process control) components. Another release will shortly add new 'recipe based' data mining approaches.
A library of data queries and analytic action scripts can be defined, providing a powerful central actions store from which users across the enterprise can (subject to individual and group access permissions) select. The components available to the system include new multivariate quality and process analysis elements, which expand the previous portfolio considerably in some rôles.
Extensive automation and centralisation of data handling are available with this system which, along with simplified dashboard style monitoring of complex multivariate processes, allow a highly flexible graded information delivery with subtle command and control. With data analytic product an ever more vital currency in industrial governance, this system provides a means to develop standardised and reliable methods independent of operator expertise. Supporting this structure is the Monitoring and Alerting Server (MAS) which allows a whole gamut of those centrally stored analytics to be simultaneously and continuously running, each one represented on a dashboard as an icon signalling its current status. Alarm actions can be automated, and clicking on an icon generates expanded information in more traditional Statistica form. After a short familiarisation this is a remarkably rich and intuitive arrangement, mimicking the natural human system of shifting attention focus very effectively.The database approach, with which StatSoft has long experience, pays off for large data volumes, and location of the data stores to which all of this is applied seems to matter little. My test installation, hosted by an academic department, took in its stride monitoring of a multiterabyte geophysical database, continuously updating on a live wide band capture stream, two continents away. On a review system with myself as the only user I wasn’t able to properly explore the collaborative aspects (the system is designed as an organic whole, not as a set of bolted together network substructure), but they worked well under simulation scenarios with files opened and shared by software 'robots'.
A word of warning against hubris. This Enterprise System needs to be taken seriously, everything being thought through before starting installation. I made the casual (and fatal) assumption that, with my Statistica familiarity, I could just pick it up without the offered help from StatSoft UK; as a direct result of my arrogance, the whole system refused to start at all so I had to remove the whole thing and start again. I definitely do not recommend my cavalier approach to others. Time spent on familiarisation is essential to reaping benefit; treat it with the respect it deserves and it will repay you in kind.
