Dimensional Modeling is a key componend of designing effective data warehouses. This is an introduction, so the most important topics are only addressed below.
Traditional transactional systems are not optimized for analytic purposes, leading to the need for a specialized structure like in data warehouses. In contrast Data Warehouses could also be seen as analytical systems. Dimensional Modeling is a design technique that organizes data for easy and efficient querying and reporting. Roughly speaking it’s a more normalized approach commonly used in transactional systems.
This leads us to the concept of fact and dimension tables and the importance of creating star schemas or snowflake schemas to organize tables for more simplicity and performance. In this normalizing approach, fact tables containing quantitative data, for performance measures. While dimension tables describing the context of the fact data, resp. measures.
To gain further performance there is also the concept of granularity. There is a leverage by choosing the appropriate level of detail for data stored in the data warehouse or at least in the published symantic layer aligned to the needs of the business useres.
Since I am always a bit petty when using the terms Business Intelligence and Data Science, there is to emphasize the business context. In this context dimensional modeling needs to be aligned with the key business processes of an organization, ensuring that the data warehouse is closely tied to the business’s analytical needs. As a conclusion, building a data warehouse is ‚Business Process Dimensional Modeling‘. A tailor-made data model reduced to what is necessary and adaptive to change.
