A modern data architecture manages data in many different forms:

• Normalized data
• Non-relational data
• Semi-structured data available from diverse sources and a range of formats
• Data Vault data
• Dimensional data in information marts for reporting/business intelligence

The variety of forms and volume of data means that data transformation is a complex and costly aspect of any modern data environment.

Previously we have shown that Fact-Based Modeling can be used as a unified modeling approach managing conceptual data semantics across all these different forms. In this talk we will focus on showing that a Fact-Based Modeling conceptual query language can be used to automatically generate transformation code from source, through a data warehouse and to information marts.

We will apply this approach in two distinct environments:

• A traditional data warehouse based on Data Vault
• A cloud-based data warehouse

We'll show that automated transformation code generation from a conceptual query language reduces cost, contains fewer errors, and supports change/agility.