Business Intelligence and Data Warehousing
Business Intelligence (BI) is a buzzword which is often heard. The topic has gone from being a small article on the back page of a newspaper to headline news in IT over the last decade. Numerous companies have proven that processing, evaluating and analyzing your own data leaves you at a significant advantage over the competition.
Challenge
Company-wide integration of data is a complex task and is often made more complicated by the following topics:
- Data inconsistency in both an application and between various systems
- Different terms for the same data or the same terms for data between systems/departments/business sectors
- Comprehension problems for operating departments/users who only need to quickly find out a few numbers and do not want to spend months exchanging specifications using the base data
- Deadlines and lack of time during development which often leads to few or missing documentation
When actively used, a data warehouse or BI solution is never complete. The market and the company with it are constantly changing. This means that changes need to be made, which leads to the next set of challenges:
- Adapting existing central modules without changing existing data
- Estimating the time needed for changes to be made to the existing system
- Extra technical uses need to be told which evaluation options are available and where extensions need to be made to the system
MID Solution
Complex topics can be controlled using abstraction. We have developed two models for this: a semantic model and a technical model.
The semantic model depicts the overview for the operating department. Facts and dimensions show which data is ready for evaluation.
Business objects can be used to reproduce how information is linked within the company and, if applicable, new dimensions and facts can be designed. If enough information still cannot be obtained from these business objects, a load layer can be used to reconstruct which data was not taken into consideration before. This enables the operating department to recreate data in a form which they understand.
The technical model consists of a logical and a physical data model, as well as a process model. The logical model is a complete view of the data architecture, from the load layer to evaluation. Dimensions and facts are created as their own model elements.
The physical model enables the model to be adapted into various evaluation environments. This means that various evaluations can be carried out using just one logical model.
The processes are described based on the data model. This is extremely easy in 80 percent of cases; the focus is on a clear description of complex transformations in the process model.
This means that the creation of data, from the load layer to evaluation of attribute levels, can be viewed at an attribute level in the technical model.
Metadata is transferred into the next step for each modeling step. Documentation is created automatically in the process. Knowledge collected in this way grows and is available quickly and directly for employees old and new.
These models mean that changes made are not only manageable but welcome.
