Discussion Board

Fact tables are completely normalized The fact table is in normalized structure as the redundant information is maintained in the dimension tables. The fact table contains foreign keys to the dimension tables. To get the textual information about a transaction (each record in the fact table), you have to join the fact table with the dimension table. Some say that fact table is in denormalized structure as it might contain the duplicate foreign keys. It would be partially correct to say denormalized. Let me get you completely right on the concepts. Generally it is the schema that is denormalized than just the fact table. Fact table ideally contains the aggregatable numeric values and reference keys to the dimension tables for slicing purposes. In some architectures, fact tables could just look like a few columns with numbers. It is the dimension tables that help you analyze those numbers. But yes, there can be situations where fact tables are populated from numerous tables and they contain a lot of columns. In that case, you can say that fact table is denormalized, but it would be great to say that the schema is denormalized than just the fact table. Remember, it is the dimensions that help you analyze facts, so a fact without a dimension is useless. Prajakta Pandit 03-15-2017 01:53 AM

Fact Tables Given answer is not correct , fact tables can be normalized as well as denormalized subject to business context. For descriptive information we detailed it with hierarchical roll-up attributes in the same table a approach to denormalized fact table. Sometime with snowflake schema unlike Star schema, the dimensions table in a snowflake schema are normalized. Khurram Khurram 10-21-2016 08:17 PM

Write your comments

*

* Email must be in the form someone@domain.com

*

*