Here's an example of another dynamic to consider when deciding how to interact with data: the data themselves.
Once upon a time I was something of an expert in an area of business intelligence called Online Analytical Processing (OLAP). The idea behind OLAP is very similar to what we’ve discussed here previously about data warehousing. But the acronym “OLAP” is often associated in the BI world with a specific type of database that isn’t like the table-based databases we’ve discussed to this point. This type of OLAP database can be loaded into a spreadsheet like Excel so that you can investigate data (the “I” in our “PRIME” model) without having to learn a programming language. These kinds of databases are very handy for dealing with aggregations of numbers. Finance and accounting people love them.
A few years ago I entered into a consulting engagement with a company in the retail industry. When I arrived on the scene they were in the middle of developing an OLAP database application for labor management. There was a problem: the subject data were primarily textual in nature, not at all suitable for analysis using the number crunching OLAP database that had been selected for the task. Apparently the labor management group had seen a presentation of the OLAP database by the finance and accounting group and decided then and there that they, too, had to have an OLAP database. Never mind that the OLAP database made little or no sense for the kind of data they were analyzing. No amount of presentation, expert testimony, or cajoling would convince them to consider an alternative. Once the budget was allocated and the money spent, there was no turning back.
In the end we were able to build the application the customer wanted by associating a count of “1” with each bit of textual data so there would be something numeric for the OLAP database engine to crunch. We code-named the resulting solution “Chitty Chitty Bang Bang” after the flying car of the 1960s Walt Disney film. And we all lived happily ever after.
No comments:
Post a Comment