I have had my first exposure to MAD Skills: New Analysis Practices for Big Data , very interesting approach , in many ways an extreme expansion of the concept for operational analytics via OLAP. The use of the “Sandbox” environment for statisticians and analyst, can also be incorporated in current BI applications using the Agile / SCRUM Approach modded with Extreme Scoping for BI(Larissa Moss).
I certainly see the advantage of enabling “deep” access to high end users , however the issues for broader use are still “Data Prep” and resolving Data Quality issues, as discussed in another thread in the Data Science LinkedIn group, also in the article the term “Data Driven” is used in the context of the “Functional” being developed, as opposed to automated reusable processes “Driven” by metadata.
Actually the entire approach seems to be geared towards being able to run one off processes and could be well served by a discussion of applying a Metdata Mart approach (Belcher Gartner) to retaining and reusing results. Also visit the Data Science Group at LinkedIN