4 of 4 people found the following review helpful
Not about analytics and not really about BI either,
This review is from: Agile Analytics: A Value-Driven Approach to Business Intelligence and Data Warehousing: Delivering the Promise of Business Intelligence (Agile Software Development) (Paperback)
As with most data professionals I wrestle with how to apply agile principles and practises on large datasets on an ongoing basis. As data warehousing and business intelligence are my particular focus I was drawn to this book. but sadly I cannot recommend it.
* It has nothing to do with analytics
* The majority of it is vanilla agile practises and nothing to do with BI/Data warehousing
* It fails to mention what must surely be the prime concern of any BI system, DATA QUALITY
* It completely misses any data profiling activity or data exploration activity
* Where BI is mentioned it is given a very shallow treatment.
* It scatters references to Inmon and Kimball but fails to weave their work into the books theme or deal with conflicts between the agile method proposed and the methods by the fathers of data warehousing/dimensional modelling.
For example, Kimball warns against building a system to produce a particular report as this produces a stovepipe solution. Model the business process correctly and this will provide the solid foundation on which the report (and reports yet to be conceived) will be built. You don't have to model the foundation for the entire house but you do have to build the foundations in a way that is robust enough and extensible enough to prevent stovepiping. The book does not address or discuss the apparent conflict between its message and Kimball's approach.
There is one useful idea in it and that is using a message based architecture for populating the data warehouse. However it then explicitly describes the use of an entity-attribute-value database to provide an agile method of building a long term system of record. The staggering awfulness of EAV as a solution, especially in the context of a data warehouse, is hard to put into words. EAV models are notoriously difficult to query effectively, have awful performance, inherently one-size-fits-all security and a plethora of other penalties. Given that BI is all about getting information OUT this makes them fundamentally unsuited to BI work.
I do have to give the author credit for mentioning the fact that the cost impacts of change are greater the lower down the technology stack you go.
In short the book reitterates accepted agile practises and mentions DW luminaries to lend itself credibility that it does not possess itself.