Data modelling is vital to get right if you want your analytics to deliver results.
Data modelling is crucial for organising and presenting your data. In addition to aiding the visual representation of your data, it also helps to ensure its consistency (naming conventions, default values, semantics, etcetera) and enforces compliance and any other business rules.
With the increasing volume and depth of data available to organisations, good data modelling has never been more important; you need a structural foundation to really get the value out of it. Snowplow Analytics co-founder Yali Sassoon describes it like this:
“Data modelling adds meaning to raw, event-level data using business logic…We might look at the data point in the context of other data points recorded with the same cookie ID, and infer an intention on the part of the user (e.g. that she was searching for a particular product) or infer something more general about the user (e.g. that she has an interest in red pants). These inferences are based on an understanding of the business and product. This understanding is something that continually evolves; thus, as we change our business logic, we change our data models.”
Want to learn more about modelling? Check out these articles!
While Sassoon is looking at it from a website analytics point of view, the principles are the same. Good data modelling improves data quality, reduces cost, and improves performance.
Three data modelling types
There are three common styles of data model that you are likely to come across:
Conceptual data models
Conceptual data models are usually used to define business concepts and rules. They may also be called domain models, which establish basic concepts and scope for stakeholders. It identifies the high-level, user view of data. It tends to break down into these components:
- Entity: a real-world thing, for example, a product or a customer
- Attribute: the characteristics of said entity, for example, the product’s price or a customer number
- Relationship: The association between two entities, for example, the sale of a product to a customer
Logical data models
Logical data models explore domain concepts and their relationships, depict entity types, the attributes describing them, and the relationships between them.
Physical data models
Physical data models provide an internal framework for how data is stored within the system. A physical data model is created using the native database language of the database management system. It usually describes data requirements for a single project or application, with clearly defined tables and columns.
Read more articles on data modelling, here!
Data modelling and business intelligence
Good data modelling enables valuable data analytics. More than one data scientist has described data modelling as the ‘architect’s blueprint’ – providing the plan and framework that allows the data analyst to filter through it to find the insights that could make a real difference to the organisation.