For this chapter, we will use six different tables available on GitHub at https://github.com/PacktPublishing/Tableau-Cookbook-for-Experienced-Professionals/tree/main/Chapter_01.

For this chapter, all the images and examples will be provided using Tableau Desktop Public Edition, version 2024.3. Everything we will cover can also be done through the web interface of Tableau Cloud or Tableau Server, but we recommend using Tableau Public Edition or Tableau Desktop to follow along.

Note

The look and feel of the Data Source page are a bit different in Tableau’s most recent version, 2025.1. Additionally, one of the topics covered in the chapter – multi-fact data models – has only been supported since version 2024.2. For this reason, we recommend updating your version of Tableau Desktop or installing Tableau Public Edition version 2024.2 or higher for this chapter. If you can’t have the latest version, keep in mind that some images...

Data models can be very simple (a single table) or more complex (dozens of tables). For this recipe, we will start with a simple data model relating two different tables together to better understand the data source page and how to create relationships and what they are.

There are two main ways to create a data model:

• Using Tableau Desktop or Tableau Desktop Public Edition
• Using the web interface from Tableau Cloud or Tableau Server

In this recipe, we will use the first option, but you can follow along using the web interface if you prefer. To create a workbook and connect to new data from the Tableau Cloud or Tableau Server web interface, check the first two steps in this guide if you don’t know how to do it: https://help.tableau.com/current/pro/desktop/en-us/getstarted_web_authoring.htm.

Getting ready

You can download the tables we will be using for this recipe in the GitHub repository shared in the Technical requirements...

In the previous recipe, we related two tables using what we called the logical layer, but Tableau’s data model allows us to use two different layers:

• The logical layer is the default view in the Data Source page. When you add tables to your model, those are related through relationships, which have been the default way of combining tables in our data model since Tableau version 2020.2.
• The physical layer allows us to combine tables using traditional joins and unions. Before 2020.2, this was the only layer available. Each logical layer can contain one or more physical tables.

Figure 1.9 – Tableau’s data model

The main benefit of having a logical and physical layer is the higher flexibility to build data models based on your needs.

In most scenarios, working with the logical layer should be enough, but if you need to perform traditional joins or unions, you’...

So far, we have seen how to build data models and data sources using one single data connection type. Our example has been with CSV files, but the same logic will apply when using data from Snowflake, Google BigQuery, Amazon Redshift, Azure SQL Database, or any other connection type.

In a lot of use cases, data might not be in the same database type or in the same file type. Very frequently, analysts need to combine data from a database with data from a CSV file; data from an Excel file with data from a CSV file; or even data from Snowflake with data from Amazon Redshift for instance.

How can we create a data model in Tableau that combines different connection types? Learning how we can build multi-connection or cross-database schemas is the main objective of this recipe.

Getting ready

For this exercise, I’ll use some of the datasets we have used already during this chapter. If you don’t have them, please go to the beginning...

In many scenarios, when creating relationships between tables, you’ll not need to do anything other than specify the related fields between them. This will guarantee you get meaningful results and, in general, a good performance.

Sometimes, with complex data models with a lot of different tables or big datasets and complex relationships, you might want to adjust how the tables are related to let Tableau know specifically how to perform those relationships and improve query performance and overall speed.

Tableau allows us to do this by manually setting the cardinality and referential integrity of each table in a relationship. It is important to keep in mind that setting these options incorrectly can also translate into wrong results during your analysis.

A best practice is to only adjust the following settings when you have a deep understanding of your data, especially of the fields you are using to relate your tables.

Getting ready...

Tableau allows us to build data models quickly and easily. However, it lacked a concrete type of data model that is common in some companies and scenarios: multi-fact data models.

You may have noticed that all the models we have seen so far have one base table. The base table is the first table we add to the data model, and every single table we add afterward must be connected to it directly or through another table.

This has been an obstacle for a long time to developing more complex data models with different base tables not connected between them, and it has been solved with Tableau version 2024.2.

Getting ready

For this recipe, we will use the same Hospital tables we used earlier in the chapter. Practice what you have learned in previous recipes to create a data model that combines Visits, Patients, Doctors, Hospitals, and Specialties, as shown in Figure 1.19.

Figure 1.19 – The starting...

One of the main benefits of Tableau is the efficiency of Tableau extracts. An extract is a snapshot of the data that Tableau gets from the original source and saves locally or on your own Tableau Server or Tableau Cloud environment.

Tableau extracts have some benefits, the two most important being improved performance and reduced workload from the original source when users run queries using the extract.

As you can imagine, Tableau extracts also have some downsides. They can take a long time to run if the original data is very big, and to date (Tableau version 2024.2), it’s not possible to create data models combining Tableau extracts.

Fortunately, Tableau has another tool called Tableau Prep, or Tableau Prep Builder, that allows us to combine different Tableau extracts and also perform additional data preparation tasks before creating our Tableau extract, similar to what you can achieve with Extract, Transform...