The quality of BI modeling directly affects the quality of the data warehouse project, so before modeling, it is necessary to have a clear definition of the structure, size, and model functions of the data warehouse.
There are many factors that affect BI data warehouse modeling, which often vary with the specific circumstances of the project. However, some principles are the same, and the implementation process of various projects needs to be considered, and some common problems that project personnel are easy to understand also need to be paid attention to.
BI Modeling Principles
1. Modeling around the business situation: business requirements are the foundation, and the data organization of the data warehouse is subject-oriented, not report-oriented, and is oriented to the subject area of business analysis, such as common sales analysis, contract balance analysis, and customer relationship. analysis, etc.
2. Ensure the consistency of data: To ensure the correctness and integrity of the logical relationship between the data, the data warehouse must realize the integration of the data and the isomorphism of the data, and the relative stability of the data, and perform real-time reading for the realization of the application. write operation.
3. Use scheduling: The data warehouse must have the ability to reflect historical changes and timely and accurate data processing. Therefore , scheduling must be used when BI modeling is incrementally updated, that is, incremental update processing of fact data tables. Before using scheduling, consider the actual amount of data and make it clear how often the data is updated. If the amount of data is large, it can be done every day, then the data can be extracted on a daily basis, or, like FineBI, the business intelligence of FanRuan , it can be updated incrementally at regular intervals; for those with little change, it can be a week or longer. If there is a slowly changing dimension, the update of the dimension table needs to be considered during scheduling, and the dimension table must be updated before updating the fact table.
4. Balance between demand and reality: Provide user-acceptable BI solutions based on business needs. When BI modeling, it is necessary to constantly balance between user needs and data source facts, not only for designers themselves, but also for enterprise business personnel. To face this reality.
common problem
1. How to start the design of the model?
The purpose of BI modeling is nothing more than to improve the management level, which is also the core meaning of the BI project. In the early stage, you must clearly understand the business requirements, business scope, etc., and clarify the enterprise's expectations for business intelligence and which topics need to be analyzed. Collaboratively analyze the customer's current management level, enterprise structure and operation process, and what are the weak points and key points of the management process to help enterprise personnel understand their own needs.
2. Implement ignore confirmation process
After confirming the needs of the enterprise, many project personnel feel that they can design and implement it radically, but in the actual process, they often encounter various problems that do not match. The reason is that there are always omissions in the preliminary discussion process, and some personnel do not have a deep understanding of the business, resulting in continuous adjustments and delays in the project cycle. Therefore, in the modeling process, it is necessary to constantly confirm the business analysis model and whether the data can support it. The implementation of a good business intelligence BI project usually fully understands the business system of the data extraction object, fully communicates with the business personnel, and repeatedly confirms with the leaders, so as to avoid the subsequent repeated work of the enterprise and increase the burden on the enterprise.