An article I’ve read time ago gave me the idea to attach an extra table to manage the variable number of columns in different data sets*. This article about dimensional modeling was introducing the concept of “helper table”. In few words, I can create a single column and store a reference (a code, an ID) to another table which holds the real dimensions, no matter how many columns they are. I simply need to create a fake code for the dimension group and link the two tables on that code.
The aim of these posts is trying to find an elegant database solution to this: a way to build a common repository for different data sets from different softwares so that data can be easily exchanged and the history of the changes can be visible to everyone involved in data analysis. And -much better- a new interface will help maintain, correct, cleanse the various data sets.
Stay tuned for upcoming articles on how to read/write panel data to an Oracle database and read them from several mathematical or statistical packages.