BG Tableau user group Q3 2024 Meeting Slides

Fall with Data
BGTUG Meeting Q3 2024

Agenda
➔What do we want to see next?
➔Relationships in Tableau
➔Trivia time!
➔Trivia debrief
➔Networking

What do we want to see?

Multi-fact Relationship Data Models
➔Allows you to build data sources with more than one base
table
➔The base tables are always the left-most tables in a data
model
➔Creating trees of tables, allows the generation of structures
with different domains and use their shared characteristics to
connect them.

Levels of relatedness
➔Data models with multiple base tables provide a lot of ﬂexibility to the developer.
➔Relatedness is only relevant for multi-fact relationship data models, prior to this
feature everything was either related (tables within a single data source) or not
related (data blending across multiple data sources)
➔In a multi-fact relationship model, tables can be:
◆ Related - prior to 2024.2, all data sources were single base table with all tables related
◆ Unrelated - those tables are in different trees. Base tables are ALWAYS unrelated to each
other. Downstream tables of exactly one table are only related to the other in the tree.
◆ Shared - have multiple incoming relationships (belong to more than 1 tree). Downstream
tables after a shared one are also considered shared.
➔Relatedness amongst a group of ﬁelds is determined on sheet-by-sheet basis. To
create a viz with ﬁelds from multiple tables, Tableau performs joins to compute values.
◆ Related ﬁelds are used, related dimensions are inner joined. NB additional joins might be
performed in order not to miss values.
◆ Unrelated ﬁelds - dimensions are cross joined, measures - table scoped. Ambiguous or not yet
related ﬁelds are treated the same way.
◆ Stitched ﬁelds - similar to linking values in data blending. Dimensions are outer joined,
measures - aggregated to a level at which could be broken and might be repeated.

Field-levels of relatedness
➔Relatedness between ﬁelds is determined on sheet-by-sheet basis taking into account
the data model structure, ﬁelds actively in use, if ﬁelds are dimensions or measures.
➔In a viz, using a dimension from a shared table stitches together otherwise unrelated
ﬁelds and allows them to be evaluated in the same viz. (ALWAYS bring out a stitching
ﬁeld before any unrelated ﬁelds). Measures can’t stitch!
➔The queries used to calculate the values shown in a viz rely on joins.
➔An immediate inner join is computed for each of the unrelated ﬁelds and the stitching
ﬁeld in turn, and then those results are outer joined based on the values in of the
stitching dimension.
➔Measure values repeat for unrelated dimensions.

When to use a multi-fact relationship
models?
➔Allows unrelated tables in the model to be analyzed as long as they share tables in
the model.
➔It should be used when the data spans different concepts (multiple fact tables of
unrelated contexts)
➔Moving through stages in the same context (applications, transcripts, alumni events
for students)
➔Different context for the same events - base tables for appointments and invoices,
sharing the context of patients and doctors
➔Domains that might correlate - sales and weather conditions through date and
location. Better done with blending.

Difference between relationships and joins
➔ Relationships are a dynamic way to combine data from multiple tables. You don’t deﬁne join
types. You only need to select matching ﬁelds to deﬁne a relationship.
➔ Relationships between tables can happen only between ﬁelds of the same data type.
➔ Relationships can’t be deﬁned based on geographic ﬁelds.
➔ Relationships between published data sources cannot be deﬁned.
➔ Relationships can be many-to-many and support full outer joins. When you combine tables
using relationships, it’s like creating a custom data source for every viz, all in a single data
source for the workbook.
➔ Relationships do not require LOD for analysis at different levels of detail.
➔ Joins are a more static way to combine data, deﬁned upfront.
➔ Joined tables are always merged in a single table.
➔ Joins may drop unmatched measure values.
➔ Joins may duplicate aggregated values when ﬁelds are at different level of detail.

Difference between relationships and data
blend
➔Both support analysis at different levels of detail.
➔Relationships - deﬁned in the data source
➔Blends - in the worksheet
➔Blends - support only left joins
➔Relationships - support full outer joins
➔Relationships are computed locally
➔Related ﬁelds are ﬁxed, but depend on the sheet when data is blended

Keep in Mind!!! Or some possible pitfalls
➔Per table ﬁlter extracts - Filtering results may be different between live and extract
connections. All extract ﬁlters are per-table.
➔Row level calculations can only refer to ﬁelds which share the same upstream base
table.
➔All ﬁelds in the combined ﬁelds share an upstream table.
➔Sets can only be created involving ﬁelds that share the same upstream base table. In a
viz, the option to Add to Set may be available from a mark when that mark is deﬁned
by ﬁelds unrelated to the ﬁelds used to deﬁne the set. If you choose Add to Set,
Tableau will add only the related ﬁelds to the set deﬁnition.
➔LOD INCLUDE - can’t be evaluated across unrelated ﬁelds. It is possible to have a
valid LOD, which is invalidated in the context of the speciﬁc viz.
➔LOD EXCLUDE - may be incorrectly marked as not valid across unrelated ﬁelds.
➔Updating published data sources - create a copy before turning it into a multi-fact
relationship model.
➔Tableau Pulse may not work with multi-fact relationship data models.

Trivia Time!

You need to change the values of a dimension without creating a new ﬁeld.
What should you do?
a) Rename the ﬁelds.
b)Create aliases.
c)Create groups.
d)Transforms the ﬁelds.
Which Tableau feature allows users to create customized groups of data points based
on speciﬁc criteria?
a) Calculated ﬁelds
b) Quick ﬁlters
c) Parameters
d) Sets

If you publish data to Tableau Public, can it be now viewed by anyone?
a)No
b)Yes

Which of the following ﬁles types contains a Tableau Workbook along with local data?
a).hyper
b).twb
c).twbx
d).tde

Which of the following is not a Tableau ﬁle type?
a).tds
b).tbm
c).tdx
d).tdsx
Which of the following Data Pane items cannot be dragged onto the Filters Shelf?
a)Dimensions
b)Measures
c)Sets
d)Parameters
How many maximum tables can you join in Tableau?
a)16
b)8
c)32
d)2

Tableau developed a product called Vizable?
a)True
b)False
You can add additional Color palettes to Tableau
a)True
b)False
When using data source Fields in Calculated Fields, the Square brackets are optional
a)True
b)False
In a Calculated Field, the following will be interpreted as a comment
a)Anything after //
b)Anything after ##
c)Anything after –
d)Anything between /** */
e)Anything between “ ”

What is the maximum number of Columns in a Table
a)10
b)12
c)14
d)16
e)18
You have the following dataset. You plan to create a dashboard that will be ﬁltered to
show only data that is relevant to a speciﬁc Tableau user based on the
Tableau_User_Name ﬁeld.
You need to create a boolean calculated ﬁeld to place on the data source ﬁlter.
Which formula should you use for the ﬁlter?

a)ISFULLNAME([Tableau_User_Name])=USERNAME()
b)[Name]=[Tableau_User_Name]
c)USERNAME()=[Tableau_User_Name]
d)[Name]=USERNAME()

You plan to create a visualization that has a dual axis chart. The dual axis chart will
contain a shape chart and a line chart. Both charts will use the same measure named
Population on the axis.
You need to configure the shapes to be much larger than the line.
What should you do?
a)Duplicate Population Drag the duplicate to the second Marks card and conﬁgure the size of the marks
independently
b)For the second axis, select Shape on the Marks card. From Select Shape Palette, select Custom, and then select
Reset
c)Create a custom shape that is larger than the default shape and add the shape to the Shapes folder in My
Repository.
d)Change Population to a discrete dimension.
You need to ensure that the data is updated every hour. What should you select?
a)Extract Refreshes
b)New
c)Connected Workbooks
d)Connections

You have a line chart on a worksheet. You want to add a comment to March 2020 as
shown in the following visualization. What should you do?
a)Drag the growth rate to Text on the Marks card.
b)Add a tooltip.
c)Annotate March 2020.
d)Enable captions.

You have the following sets in a Tableau workbook: Top N Customers; Customers of
2020; Top N Products; Sellers of 2020. Which two sets can you combine? (Choose
two.)
a)Top N Customers
b)Sellers of 2020
c)Customers of 2020
d)Top N Products

You have the following dataset. Which Level of Detail (LOD) expression should you use
to calculate the grand total of all the regions?
a) { FIXED [Region]: SUM(Sales) }
b){ FIXED : SUM(Sales) }
c){ FIXED [Region]: TOTAL(Sales) }
d){ FIXED : TOTAL(Sales) }
You have the following dashboard. Which two elements are ﬂoating? (Choose two.)
a)The title
b)The color legend
c)The map
d)The Rate of Obesity chart
e)The state ﬁlter

You are the owner of an alert. You receive an email notification that the alert was
suspended. From where can you resume the suspended alert?
a)The My Content area of Tableau web pages
b)The Data Source page of Tableau Desktop
c)The Notiﬁcations area of Tableau Prep
d)The Shared with Me page

Trivia Debrief