What are the characteristics that make a skilled data modeler?
In short, it always helps not only to know technology well, but also to have experience in both working with data and understanding business side of information stored.
So the first thing that an aspiring modeler needs to do is to stay very close to the business, understand the products, their life cycles, purpose, why they exist.
Ask a lot of questions if you have no product knowledge, otherwise you will build a sound technical model that doesn’t work for business needs.
Spend time being immersed in the daily business, learn about the problems they are facing, talk to all levels of the organization.
Second: test the model by trying to use it, build an ETL prototype, mock up some data and ask users to validate the model. Test all relationships, and iterate through this modeling process.
This allows you to discover defects much sooner, before they become very costly to repair – for example when the development team has started using the model.
Third: always challenge your assumptions. It’s a weak well-known psychological trap, we as human beings, always look to confirm our opinion or idea.
But the right thing to do is to challenge it, look to disprove it, and if you can’t then you know the assumption passes the test. Also, always test the model, the sooner the better – as you build it incrementally – see if it’s able to solve both business and technical problems the organization is facing.
Fourth: pay attention to the API side of the model, how well it integrates with the rest of organization data, systems, and other models
Fifth: watch and observe, if entities look similar, try to generalize them. For example, adding new categorization attributes may allow you to reduce the number of entities if they turn out to be from the same category.
Sixth: stick with the project through the implementation, you will help the development team to work with the data and to fix any issues in the model should they arise.
Happy modelling!