DATA MODEL
• What is a Data Model?
• In short, a data model is the collection of
• loaded tables of data
• relationships between the loaded tables
• the measures (formulas) written that apply the business rules to the raw
data to extract business insights (e.g., sell price minus cost price is
profit), including
• the formulas themselves
• the formatting applied to the formulas
• the business names given to the formulas
• Power BI is a data model based BI reporting tool; not all BI tools
are data model based. One example of a non-data model based BI tool
is SQL Server Reporting Services (SSRS). SSRS is a non-data model
based BI tool.
• Non-data model based BI reporting tools require the report writer to
first generate a query to fetch the data from a database (typically SQL
Server) and return the results of that query to SSRS so the results can
be rendered in a report. Many non-data model based reporting tools
have a user interface that helps with the generation of the query (if
ONE want to use it), or ONE can also use a scripting language like T-
SQL to fetch the data that ONE need for each report.
• A second example is Excel. Traditional Excel is also a non-data
model based BI tool. By traditional Excel, I mean Excel the
spreadsheet tool without the modern BI add-ins of Power Query and
Power Pivot. In the case of traditional Excel, the user loads the data
into the spreadsheet and then logically relates and aggregates the data
using Excel formulas and builds a report (often on a new sheet) to
summarise and present the results (the report).
• The benefits of a data model based tool like Power BI are:
• the author of “the data model” builds a reusable capability (the data model)
that can be used to solve the current reporting requirements as well as
(often) future requirements without the need to go back and write further
queries to retrieve a new subset of data.
• The author can often be a business user (normally with good Excel skills)
– they don’t have to be a professionally trained database administrator or
SQL professional.
• The model is “conceptual” in nature, supported by a user friendly interface
that let’s ONE “build” the model logically with the minimum of coding.
Note: ONE will need to do some coding (writing formulas) for a good model,
Relationship Properties
• A model relationship relates one column in a table to another
Column in a different.
Cardinality
• Each model relationship must be defined with cardinality.
• There are four cardinality types.
• “from” and “to” related columns.
• “one” side means the column contains unique values.
• “many” side means the column can contain duplicate values.
• One to many
• Many to one
• One to one
• Many to many
• The designer automatically detects and fixes the cardinal type.
• It queries the model to know the column that have unique values.
One to many / Many to one
• Most common cardinality type.
• Choosing the order of relationship between tables.
• ProductId(Product) – unique values
• Salesamount(Sales(productid)
One to one
• It means both columns(from different tables) contain unique values.
Many to many
• It means both the columns(from different tables) contain duplicate
values.
Cross Filter Direction
• Each model must be defined with cross filter direction.
• The possible cross filter options are dependent on the cardinality
type.
c
• Single cross filter direction means single direction.
• Both means both directions.
• Relationship that filters in both directions is called bi-directional.
• One to many relationships the cross filter direction is from the one
side.
• One to one relationships – direction is always from both tables.
• Many to many relationships – Single/Both
• In Power Bi Desktop Model view, a single arrowhead represents a
single direction.
• A double arrowhead represents a bi directional relationship.
