Modelling and design of data warehouses

OData support
Dr. Martinek Péter
Department of Electronics Technology

The drive of the emergence and continuous improvement of data warehousing is that it is indispensable for companies. The intensive and turbulent competition on the market requires analysis of the tremendous amount of data available. Better and better data quality as well as quantity showing a correlation between variables means advantage in the competition and business benefits.

Data warehouses are complex systems which contain a company’s full range of data in time series while satisfying the various demands generated by the different business intelligence systems. In parallel with their improvement several methodologies were born to attempt modelling data warehousing, but neither of them become popular.

In the first part of my thesis work I focused on reviewing the different designing and modelling systems. I got to the conclusion that most of them are concentrating on one subtask of data warehousing, and there is only a handful of them which attempts to cover the entire system.

My next task was to choose a methodology; following this methodology I designed a small data warehouse for an imaginary bank, which stores the data of credit portfolios of resident customers. For that I had to get familiar with basic definitions and the structure of banking, which later on I could use to create the source systems and then using the PL/SQL language to generate the data. The chosen method was the so called SOA data warehouse designing. This method uses the UML modelling language which starts with the input requirements and drives towards the more and more detailed design. I got to the conclusion regarding the methodology that it is well suited to practical use, however there are some defects, for which I made recommendations to improve such as creating an UML profile specifically designed for this method or creating a model for the access handling.

At the end I implemented the data warehouse I built. As first step I created the structure of the data warehouse, than with the help of Oracle Data Integrator I implemented the mapping of the ETL processes (Extract-Transform-Load). At the end using the mapping I have runned a data upload, than I created a couple of reports of credit portfolios on the Oracle Application Express platform.


Please sign in to download the files of this thesis.