Data prep techniques
Some of the data, as in any real-world case was messy...
Whatever the raw datasets would be, I had to ensure I tailored data so that I could have a star-schema. This is the best solution since it's scalable and performant for the Power BI tool.
Let's look at the Accounts dataset.
Since some of the data was missing in certain foreign keys (e.g. the client SAP id field in order to connect to transactional views from SAP BW S4 dataflows), in order to ensure referential integrity, I had to fill the missing values with unique identifiers.
Another good example of data prep is the Opportunities dataset. The CRM's data model is a snow-flake style, which isn't the best option for Power BI. In order to ensure that my model maintained it's star-schema, I had to join multiple tables (including junction tables from the system), so that I could link the Opportunities with the Accounts and Product dataset.
Diving more deeper into the realms of debt and credit data where no one even had an ERD, I also had to create composite keys which were a concatenation of various functionally relevant fields. I also had to do some error handling since some of the data had null values.