The Intelligent Enterprise magazine has a substantive RFID/ data analytics article written by Rajeev Kasturi coming out in the September 1st issue.

Entitled "Beyond Quick and Dirty RFID" it is already online and offers a detailed look at how companies are using RFID for better forecasting and tracking, and reduced shrinkage and stock-outs:

Because it's real time, location aware, state aware, time aware and highly granular, RFID data is inherently rich in properties. This property set offers a prospector's paradise for data mining and analytics. BI [business intelligence] derived from RFID data can be applied in various facets of the enterprise, from operations refinement and proactive planning to executive decision-support systems. And RFID data mining brings better insight to product velocity, receipt, storage and transportation analytics, trend analyses and exception reporting applications. On the cutting edge are pattern recognition and complex event processing (CEP) algorithms that will examine events, time and exceptions to infer meaningful patterns. This could help organizations spot promotions that would boost sales of an item if packaged with an accessory or determine that a specific shipping route is prone to delays.

Use of BI is best planned in advance rather than cooked up as an afterthought. Unfortunately, most companies grapple with implementation and back-end integration first. Once the volumes of data build, companies begin to assess BI development and tools.

What I found especially helpful about this particular article is the way it begins to sort out the messy and evolving RFID landscape that we are currently in. The landscape right now can be pretty baffling for "newbies" trying to get their arms around it.

For instance, the article sorts out the strengths of the various middleware vendors, as shown in the image accompanying this post (click on it for a larger image).

This one is a must-read.