What is data modelling?

Broadly speaking, data modelling is a way of mapping how data flows and connects. It’s a method to keep data organized, logical, and consistent. More specifically, data modelling is how Data Analysts prepare company data for their analysis and other uses. By joining data together and structuring it to make it usable.

The 3 stages of data modelling

  1. Conceptual: A conceptual model is a high-level schema of the requirements for the data model. It’s a way of scoping the next stages of the model’s development and outlining the ultimate goals of the data model.
  2. Logical: From the conceptual model, a logical model is created that brings a greater level of detail to the schema. The logical model will outline the structure, destination, and other specifications of the data model, but will be Database Management System (DBMS) agnostic.
  3. Physical: Finally, a physical model is produced from the logical model. This stage is when the data model is truly actualized and defined in the context of the DBMS. It will indicate the dependencies of and the relationships between tables relevant to the model.

To summarize it briefly:

  1. Conceptual modelling: What do I want my data to look like?
  2. Logical modelling: How do I want to be able to interact with my data?
  3. Physical modelling: how do I actually query my data?

What is data modelling used for?

Data modelling both enables and improves the overall data operations at an organization. It’s a key step towards making business data usable and achieving data-driven business. Without data modelling, your data will sit inert in your various data sources or your data warehouse. The main ways to activate your data models are reporting, data visualization, reverse-ETL, and automation.

  • Reporting and data visualization: Your data models will feed various business dashboards to track KPIs and core business metrics. Not only that, but your data models will serve as clear technical definitions of those core metrics. This will help create consistency across departments and avoid data silos from developing.
  • Reverse-ETL and automation: You can activate your data using reverse-ETL pipelines that will rely on your data models. This will bring data out of your warehouse and into the various tools your teams use on a daily basis. Once clear, reliable data is brought into your apps and software, you can use it to automate tasks and workflows.

Weld’s data modelling tool

Weld’s data modelling tool makes creating and storing your models simpler and more straightforward than ever. With smart autocomplete, code folding, error highlighting, audit logs, and version control, keeping all of your data models organized is easy. And, you can quickly see the bigger picture of your models’ dependencies, monitor their performance, and troubleshoot any issues with the lineage and observability features.