Maven Advanced Tableau

FOR BUSINESS INTELLIGENCE
With Award-Winning Tableau Expert Dustin Cabral
*Copyright Maven Analytics, LLC

Course Structure
This is a project-based course, for students looking for a practical, hands-on, and highly
engaging approach to learning business intelligence with Tableau Desktop
Additional resources include:
Downloadable Ebook to serve as a helpful reference when you’re offline or on the go
Quizzes & Homework Exercises to reinforce key concepts, with step-by-step solutions
Bonus Projects to test your abilities and apply the skills developed throughout the course

Course Outline
1 Table Relationships Explore Tableau Relationships and data modelling tools

inside the data source interface, sheets and visualizations
Create dynamic dashboard views leveraging multiple

2 Dynamic Design dimensions and measures, sheet selectors, container menus,
background templates, set/parameter actions and more
Develop custom mapping solutions including buffer

3 Geospatial Mapping calculations, wms maps, map layers, hex maps, and geospatial
fields
Leverage complex calculations to deliver actionable insights for

4 Advanced Calculations specific business use cases
Use native and integrated predictive tools in Tableau including

5 Predictive Analytics std deviations, forecasting, clustering, and R/Python integration

Introducing the Course Project
THE You’ve just been hired by Maven Roasters, a bespoke, small batch coffee roasting company.
SITUATION Your role? Use advanced analytics and visual design to take Maven’s BI reports to the
next level!
THE Maven Roasters needs a way to track KPIs (sales, profit, units, returns), compare
BRIEF performance across markets, analyze category profitability, and identify high-value
customers.
All you’ve been given is a folder of excel files containing information about orders, returns,
products, customers and territories.
THE Use Tableau Desktop to:

OBJECTIVE • Create relationships between data sources
• Design dynamic dashboards & visuals
• Leverage geospatial mapping features
• Apply advanced calculations to key business use cases
• Integrate predictive analytics assets

Setting Expectations
1 This course is designed to cover advanced topics in Tableau Desktop

• We strongly recommend completing our Up & Running with Tableau Desktop course first
• This course dives deeper into advanced filtering, blending, set/parameter actions, mapping, predictive
analytics and dynamic visual design, but not all Tableau Desktop features and integrations will be covered
2 What you see on your screen may not always match mine
• Tableau desktop updates on a monthly basis for minor releases and quarterly/yearly for major
releases, so features and functionality may change over time
3 This course is primarily geared towards data analysis and visualization

• Data preparation is another key component of the analytics and business intelligence workflow, but we
will cover this topic in depth in a separate course (Tableau Prep)
4 We will not cover Tableau Server/Online as part of this course

• This course will focus on Tableau Desktop specifically; online sharing and collaboration features will
be covered in depth in a separate course (Tableau Server)

Relationships
Relationships are part of Tableau’s dynamic Data Model interface, which automatically assigns
join conditions between related tables without materializing those tables into a single data
source
Reduce Data More Use Cases Trust Your
Preparation Per Source Results
• Reduce up front data prep by • Reduce the number of data • Retain measures that could be
utilizing flexible relationship sources pulled in to answer lost or distorted through joins
joins analytic questions and blends
• Relationships push the join to • Preserve all detailed records • Relationships ensure measures
the viz-level, allowing for from a relationship source with will not duplicate or
many more possible join multiple tables overstate due to join clauses
scenarios to occur from the
same source

Tableau Data Model
Tableau Desktop’s Data Model is split into two distinct layers: Logical & Physical
• Both layers can be used in conjunction with each other
• Source data should be clearly understood prior to enabling either data model layer; however, the logical layer is
generally more forgiving than the physical layer
Logical Layer represents the canvas for

creating relationships between tables
Physical Layer represents the canvas for

creating unions & joins between tables
NOTE: The Physical layer is the same data model structure as the
original data model prior to 2020.2.
The Physical Layer physically joins tables together,

while the Logical Layer keeps tables separate, but
defines relationships between them

Logical and Physical Layers
Logical Layer Physical Layer
• Tables dragged here are called “Logical” tables • Tables dragged here are called “Physical” tables
• Logical tables are containers for “Physical” tables • Physical tables are merged into a single flat table that defines the
logical table.
• Logical tables remain distinct, not merged in the data
source • Joins / Unions connect physical tables together via related fields
• “Noodles” connect Logical tables together via related fields
PRO TIP: To access the data model view from pre-2020.2 (i.e. joins, unions,
Venn Diagrams), simply double-click the table brought into the data
canvas

Relationships vs. Joins
Relationships Joins
• Logical tables are displayed using “noodles” to • Physical tables are displayed using Venn Diagrams
identify relationships between tables to identify joins between tables.
• Require matching fields between logical tables • Require join clause & join type selection
• Keeps each table separate, and only performs • Joins data into one physical table with specified
joins at viz-level when needed granularity defined
• Avoids data duplication through smart • Can support row-level security and extract filters
aggregation

Requirements for Relationships
Dirty Data can make relationships more

complex to use. Well-modelled data is
preferred when using relationships.
Avoid upfront data source filters
when using relationships. These limit
automated Tableau join-culling.
Matching fields must share the same data

type. Types cannot be changed in the data
interface (must be done at the database
level) Tables with many unmatched values
and models with multiple fact tables
and with multiple dimension tables can
make relationships less intuitive.
Geographic fields cannot be used to
define relationships (use the physical
layer for these types of joins)
Creating Relationships
Drag logical tables into the canvas:
Define matching fields

between logical tables:
Set Performance Options

(only change if needed):
Relationships are not supported for data source connections including Cube Sources, Stored
Procedures, Splunk, JSON, Salesforce (single source only), and SAP HANA (with OLAP chosen)

Optimizing Performance
Performance options are settings that define cardinality (uniqueness) and referential integrity
(matching records) between two tables within a relationship
Cardinality
Determines if tableau aggregates before or after joining data during
analysis
Many - if field values aren’t unique or if you aren’t sure. Aggregation will
occur before Tableau joins.
One - if field values are unique. Aggregation will occur after Tableau joins.
NOTE: If “One” is chosen and records are NOT unique (including multiple null records),
duplicate values will occur
Referential Integrity
Determines the type of join used to get dimension values for a measure during
analysis
Some Records Match – if some values in the selected fields don’t match or you
aren’t sure. Tableau uses Outer Joins, and all measures will show (even unmatched
measures)
All Records Match – if all values in selected fields have a match in both tables.
Joins are optimized with Inner Joins, but records can be removed/missing due to
unmatched values.
NOTE: “All records match” treats records as if there are no null records in matching fields.
Interface Differences (2020.2)
DATA SOURCE INTERFACE DATA GRID
Pre-2020.2 Pre-2020.2
Physical Layer (Joins & Unions) Shows all records in combined table resulting from join.
2020.2+ 2020.2+
Logical Layer (Relationships) Physical Layer (Joins & Unions) Shows all records for the individual table selected in the relationship.

Interface Differences (2020.2)
DATA PANE VIEW DATA # OF RECORDS
Pre-2020.2 2020.2+
Single Dimension & Dimension & Measure Pre-2020.2 Pre-2020.2
Measure section section per table Single Combined Source View Data Preview Number of Records = Source Record Count
2020.2+ 2020.2+
Individual View Data Preview by Table TableName(Count) = Source Record Count

The 8 R’s of Relationships
The 8 R’s of Relationships help to describe the behavior caused by different relationship types
• We’ll review each of these in the context of contextual joins & smart aggregations (more on that soon!)
Contextual Joins Smart Aggregations

Full domains remain for dimensions from a single Replicate measures across lower levels of
Remain table Replicate detail in the viz only
Relevant dimension domains are shown across Resolve aggregations to the measure’s
Relevant tables Resolve native level of detail
Retain All measure values are retained
Unmatched dimensions can be recovered by using

Recover measures
Unmatched measure values are represented as zero

Represent using ZN()
Unmatched dimension/measure values can be

Remove removed with filters

Contextual Joins
Contextual joins are created and applied at the viz-level, and dynamically change the join type
based on the combination of dimensions and measures selected
Left Join (“Remain”) Inner Join (“Relevant”)

Contextual Joins | Remain
If dimensions in the view are from a single table, Tableau will show all dimension values even if no
Remain matches exist in the measure tables (think left join)
Relationships leveraged in the view
Store ID = Store ID
Sales by Store [Sales] Store Lookup [Neighborhood]

Contextual Joins | Relevant
Combining dimensions across tables displays only records where matching fields are found between the
Relevant tables (think inner join)
Store ID = Store ID
Store Lookup Customer Lookup

[Neighborhood] [Home Store]

Contextual Joins | Retain
All records from measure tables are ALWAYS retained. Normally pulling dimensions from different
Retain tables creates an inner join, but adding a measure field will show ALL measure values from that table
(think left join)
Store ID = Store ID
Store Lookup [Neighborhood]
Sales by Store [Count()]
Customer Lookup [Home Store]

Customer ID = Customer ID

Contextual Joins | Recovered
Recover unmatched dimensions by bringing in the respective “Count” field. In the below case, bringing
Recovered in CNT(Store Lookup) allows us to view Neighborhoods without Home Stores.
Store ID = Store ID
Sales by Store [Count()] Store Lookup [Count()]
Customer Lookup [Home Store]

Customer ID = Customer ID

Contextual Joins | Represented
Unmatched measure fields (excluding the default source Count fields) will display as null by default,
Represented but can be displayed as zero using the ZN() function
Store ID = Store ID
Store Lookup [Store Type]

Sales by Store
[Quantity Sold]

Contextual Joins | Removed
Use the “Count” field from the table you want to remove nulls from and choose > 1 to remove nulls
Removed without removing ALL nulls (such as Neighborhoods without a home store)
Store ID = Store ID
Sales by Store
Store Lookup [Store Type]
[Quantity Sold]
Sales by Store [Count()] Store Lookup [Neighborhood]

Smart Aggregation
Smart aggregation allows measures to aggregate at the level of detail of their pre-join source
Physical Layer Joins: Relationships:

Left Outer Join on Customer ID and Product ID Matching fields on Customer ID and Product ID
Sales by Store Employee Lookup Employee Lookup

Sales by Store [Sales]
[No Field]
Food Inventory Sales by Store [Store ID] Food Inventory

[No Field]

Smart Aggregation | Replicate
Replicate Replicate measures across lower levels of detail in the viz only
Store Lookup
Store ID = Store ID [Neighborhood]
Customer Lookup
Sales by Store [No Field]
[Gender]
Customer ID = Customer ID Food Inventory

[Count()]
With a join, the many-to-many relationship between the Store Lookup and Customer Lookup tables would multiply
Food Inventory values to incorrect proportions. Relationships replicate across measures in the view.

Smart Aggregation | Resolve
Resolve Resolve aggregations to the measure’s native level of detail
Store Lookup
Store ID = Store ID [Neighborhood]
Customer Lookup
Sales by Store [No Field]
[Gender]
Customer ID = Customer ID Food Inventory

[Count()]
Subtotals resolve to the measure’s native aggregation (Food Inventory table) and do not aggregate replicated values across
dimensions.

HOMEWORK: Relationships
THE The manager at your favorite bookstore, Coffee & Books, has enlisted your help to bring
SITUATION their data together. You’ll need to use table relationships to create a data model for
the team to use to support their business.
THE The team needs you to create a dynamic data model for their team to accurately analyze
BRIEF sales, profits, and product mix. The key will be to create a data source that is extremely
flexible, but also provides peace of mind that the values in the reports are correct!

OBJECTIVE • Connect & transform raw data
• Relate source tables via matching fields
• Provide starter example analysis sheets for the team

Dynamic Design
Dynamic Design covers areas of dashboard creation that fall beyond basic sheet construction and
interaction used in standard dashboards
• To make your dashboard design stand out, focus on visual appeal, flexibility, and interactivity
Visual Appeal Flexibility Interactivity
• Elevate the visual appeal of your • Swap visuals, measures and • Leverage set & parameter
dashboard using external design dimensions to facilitate flexible actions to enable deeper analysis
tools and dynamic visual analysis and keep users and app-like interaction with
techniques engaged your visualizations

Visual Appeal
Visual Appeal creates an engaging platform for users to explore and interact with your dashboard
• While analytics should be the backbone of your work, visual appeal can drive user adoption as well as
comprehension of your vision
Background Templates: Information Indicators:
Background templates provide visual frameworks Information indicators come in many forms including
for developers to build dashboards. Templates are KPI shapes, Info Icons, Instructions/Commentary
both creative and practical tools for design.
These devices communicate complementary data to
Note: Templates do not refer to “Chart Templates”, but rather better inform consumers of dashboards
visual guides to overall dashboard design

Background Templates
Background Templates provide visual structure for developers to build stunning dashboards
• Templates can speed up development time and increase visual appeal of your analytic tools
PowerPoint
Create background in
PPT, save as image file
NOTE: 1” in PPT = 100

Combined Dashboard
Pixels in Tableau Template image floated
behind tableau sheets,
Sheets are floated inside
containers.
Tableau Sheets
Sheets are created
and floated on the
dashboard pane
PRO TIP: Create background templates for each device (desktop, phone,
tablet) and bring into your default view for dynamic template usage.

KPI Shapes
KPI Shapes provide visual indicators that convey complementary information to numeric values
• KPI shapes can be created via shapes, images or Unicode characters (depending on the use case)
Unicode Characters Calculated Fields Text Labels

1 Copy Unicode character shapes from free
2 Create conditional calculated fields with
3 Layer measures, calculated fields and
online web-pages Unicode characters embedded in text. other text values as dynamic sheet labels
https://en.wikipedia.org/wiki/Geometric_Shapes

Info Icons
Info Icons can be used to provide detailed complimentary information within a dashboard
• Info icons are typically created using a combination of custom shapes and tooltips
My Tableau Repository Edit Shape Menu Tooltips

1 Download free icons and place in your
2 In your sheet, open the shape menu and
3 Type details into your tooltip to provide additional
shapes folder (My Tableau Repository > choose your custom icon. information for your users.
Shapes)
https://icons8.com/icons

Instructions & Commentary
Instructions & Commentary provide end-users with helpful reference points to better utilize and
understand your dashboard
Normal Dashboard View: Instruction & Commentary Overlay:
1 Container Button
Use containers with buttons to enable
toggling of an instruction panel.
2 Custom Image
Create an overlay image (in PPT)
that is semi-transparent and
embedded in a container.

Flexibility
Flexibility provides developers the ability to create layered analytics within a dashboard, both in the
visualizations themselves and supporting filtering and complimentary content
• Two fundamental concepts in flexibility are field swapping and containers
Field Swapping: Containers:
Dimension/Measure 1
Dimension/Measure 2
Field swapping allows users to dynamically change Containers allow users to swap out entire visuals
the fields being displayed INSIDE a visualization or reveal additional filters or information

Multi-Dimensions
Multi-Dimensions enable users to dynamically swap out dimensions used in Tableau visualizations,
and can be created in 3 steps:
1 Parameter 2 Calculated Field 3 Visualization
Create a parameter to allow users to Create a calculated field using a case statement Create a visualization using the dimension
select desired dimension to display driven by the previously created parameter calculated field and selector
NOTE: Use integer for faster performance NOTE: While IF statements can also be used, CASE
and a text display label statements are best for this use case

Multi-Measures
Multi-Measures enable users to dynamically swap out measures used in Tableau visualizations,
and can be created in 3 steps:
1 Parameter 2 Calculated Field 3 Visualization
Create a parameter to allow users to Create a calculated field using a case statement Create a visualization using the multi-
select desired measure to display driven by the previously created parameter measure calculated field and selector
NOTE: Use integer for faster performance NOTE: Aggregation must be applied at this level if differs
and a text display label by measure

PRO TIP: Dynamic Formatting
Multi-Measures add complexity due to mixed formatting (i.e. swapping between Sales and Profit %)
• Additional multi-select labels and aggregation calculations can be used to customize formats
Pre & Post Labels:
Leverage the multi-measure parameter in two calculated fields, one for

before (such as a $) the value and one for after (%, K, M, etc.).
Dynamic Aggregation:
Leverage the multi-measure parameter in a calculated field used for measure

label display. This will allow for different levels of format aggregation per
measure.
Sheet Selectors
Sheet selectors allow developers to change visuals within the same physical space in a dashboard
• These are especially useful to maintain user engagement and keep dashboard space clean and focused
Outer Container
Used to house multiple
sheets within Parameter Selector
Parameter drives a calculated
field placed on filter shelf – if
sheet has no data it shrinks.
Compressed Sheet
Sheet is shrunk to 1px width when
parameter driven filter is applied.

Filter Menus
Filter Menus enable users to show/hide filters on a dashboard
• Using filter menus helps conserve dashboard space, enhance usability and create app-like experiences
1 Show Button 2 Customize Settings

Dashboard Items to Show/Hide
Select the container (by name) to show/hide in the dashboard
Button Style
Text or Image are options. Text will display flat text, while image
will all for custom icons
Button Appearance
Text/Icons need to be chosen for “shown” and “hidden” states
Image
Allows you to choose custom images/icons (default options are
menu stack and X)
Select container and choose “Add
Border/Background
Show/Hide Button”
Add a border color or background color to the button
NOTE: Content (sheets, images, text, etc.)
needs to be added to a container for this Tooltip
option to show Flat text displayed when hovering over the button

Interactivity
Interactivity grants users the ability to dynamically change visualizations, their analytic components
and physical structure simply by selecting a visualization
• Parameter & Set Actions enable next-level dashboard interaction for many types of use cases:
Multi-Measure Index Chart Year-Over-Year Sheet Selector
Relative Dimensional Proportional Asymmetric Color Scaling

Analysis Brushing Drilldown

Parameter Actions
Parameter Actions allow end-users to apply parameter values by selecting a visualization
• These actions drive the flow of analysis in an intuitive manner without drop-down box functionality
Name
1 Create Parameter Action 2 Customize Settings Choose parameter action name
Source Sheets
Select sheets to where action will be
generated from
Run action on
Choose to action on Hover, Select, or Menu
Target Parameter
Select parameter to be impacted by the
action
Value
Select field from source sheet to push to
parameter
Create a parameter action from the dashboard > actions
menu (shown as “Change Parameter” in the action menu) Aggregation
Select an aggregation type for field (if
NOTE: Parameter actions cannot be created from a sheet directly as applicable)
filter actions can be

Parameter Actions | Multi-Measure
Multi-Measures allow users to dynamically select metrics in a dashboard
• Enabling parameter actions as the driver (rather than a drop-down) creates an intuitive/immersive
experience for end-users
Calculated Field
1 Match the field value to the value of the
multi-metric parameter (i.e. 1 = Sales)
Source Sheet
2 Place calculated field on source sheet. Value
will be pushed to parameter via action
Parameter Action
3 Action will pull calc field from source
sheet and push to parameter

Parameter Actions | Index Chart
Index Charts allow for dynamic relative time comparison along an axis
• With use of parameter actions, once a date is selected all other points change relative to that date
Multi-Metric Table Calculation

Metric calc uses difference from table calc,
while referencing the selected “index” date
Multi-Metric
Normal multi-metric is dual axis to show
true value against difference from
Index Date
Selected date that is pushed by a
parameter action into a parameter
Color Legend
Sales metric is applied to color to
maintain primary metric coloration

Parameter Actions | Year-Over-Year
Year-Over-Year charts allow for trend views with dynamic date-vs-date value comparisons
• Parameter actions make this dynamically selectable from the view by selecting two points in time
1 Date Parameters 2 Parameter Actions 3 Calculations

Create both a min and max date Create both a min and max parameter Calculations will accept the values for dates pushed
parameter. These will be populated by the action with matching aggregation types into the max and min parameters and display the
comparison dates chosen in the chart (max date = max aggregation, etc.) difference in value (or percentage change)

Parameter Actions | Sheet Selector
Sheet selectors allow developers to change visuals within the same physical space in a dashboard
• Using parameter actions can make sheet selectors easier to use and more aesthetically pleasing
Parameter Action Selector

Parameter action drives a parameter paired calculated field
placed on filter shelf – if sheet has no data it shrinks
Outer Container
Used to house multiple
sheets within
Compressed Sheet
Sheet is shrunk to 1px width when
parameter action driven filter is applied.

Parameter Actions | Relative Dimensional Analysis
Relative Dimensional Analysis gives users the ability to compare a single dimension in context to
(relative to) all other dimensional values within the selected field
1 Dimension Parameter 2 Parameter Action 3 Calculation

Create a parameter to accept dimensions Create parameter action to push Calculation will accept the dimensional values from
from dimensional parameter action selected dimension to dimension dimension parameters and compare value to every other
parameter, linked to a calculated field dimensional value in the viz (shown via color scale)

Set Actions
Set Actions allow end-users to change set values by simply selecting a visualization
• Dynamic set adjustments allow for on-the-fly analysis and data exploration
Name
1 Create Set Action 2 Customize Settings Choose set action name
Source Sheets
Select sheets to where action will be
generated from
Run action on
Choose to action on Hover, Select, or Menu
Target Set | Data Source / Set

Select data source/set to be impacted by
the action
Running the action will:

Choose what the action does – assign,
add, or remove values from a set
Create a set action from the dashboard > actions menu (shown as
“Change Set Values”) Clearing the selection will:
NOTE: Set actions cannot be created from a sheet directly like filter actions Choose what clearing the selection does
– keeps, adds, or removes values from the
set

Set Actions | Proportional Brushing
Proportional Brushing updates sets using set actions by selecting dimension(s) across a
visualization on the color mark
• This shows users what portion of the overall value is represented by the selected dimension
Set Action Custom Icons

Set action drives dimensions that end up in Shapes applied to dimensions
the set. Once selected, dimensions are added create the icons for user selection
Set Applied to Color Mark

Set will inherit set action dimension
and apply to the color mark. In/out
colors will differentiate set status

Set Actions | Asymmetric Drilldown
Asymmetric Drilldown provides hierarchy drilldown for a selected dimension, while leaving other
dimensions compressed
1
Set & Set Action
Set is created from the higher- 2 Calculation
level hierarchy dimension, which The calculation defaults to blank when the
is subsequently pushed by the set is not engaged (by default the set action
set action when selected returns to empty)
When engaged, the sub-dimension appears

Set Actions | Color Scaling
Color Scaling enables cross-dimensional comparison of subsets of data which are normally
obscured by outliers within the data (i.e. very large/small values which impact scale)
Create Set
1 Create base set which will
accept values from set action
Calculated Field
2 Calculated field will show metric only
when set dimensions are enabled
Set Action
3 Set action will push selected dimensions to
calculated field (which is applied to color)

PRO TIP: Viz Animations
Viz Animations provide a dynamic transition setting for visualizations using filters, filter actions,
calculations and more
Workbook Default
By default animations are turned off (but
can be turned on at a workbook level)
Duration
Duration can be set from fast (0.3s) to
slow (2.0s) or even a custom value
Style
Style can be simultaneous (at the same
time) or sequential (in order)
Selected Sheet
Animations can be customized at a sheet
level for sheets selected

HOMEWORK: Dynamic Design
THE To better navigate the analytics career market, you’ve decided to utilize your data
SITUATION visualization skills to create a custom analytics career dashboard.
THE Your goal is to develop a robust dashboard that allows you to quickly swap fields and visuals,
apply custom-focused dimensional comparisons, and enable fast filtering and interaction.
BRIEF
Everything will need to be both functional and beautiful – this dashboard will serve as both an
analytics tool as well as a personal portfolio piece!

OBJECTIVE • Import custom templates
• Develop KPI indicators and App-like Filtering
• Enable dynamic field/visualization swapping
• Leverage Set/Parameter actions for custom business insights

Geospatial Mapping
Geospatial mapping includes any geographical visualization or calculation leveraged by Tableau
• We’ll cover geospatial mapping in the context of Data & Workspace, Use-Cases and Customization
Data & Workspace Use Cases Customization
• Ingesting geospatial data • Exploring various business • Customizing maps and

and exploring the geospatial scenarios and use cases for expanding on simple visuals
tools in Tableau Desktop geospatial mapping using internal and external
tools

Geospatial Data
Geospatial data comes in two primary forms: geographic fields and spatial data files
Geographic Fields Spatial Data Files

Can be assigned to fields in any Spatial data files are special file
source so long as field types and types that contain geospatial data
roles are assigned correctly using geometry fields (examples
include Shape, GeoJSON, etc.)
Location fields are recognized by
Tableau automatically (Country,
State, City, Zip Code, etc.)
PRO TIP: Latitude and Longitude fields are automatically generated

with standard location fields!

The Mapping Workspace
The mapping workspace provides an array of customization and feature enablement tools for
geospatial visualizations in Tableau
Map Menu
Provides additional map capability including background images and map layers
View Toolbar
Enables map search, panning,
zoom, and multiple mark selection
Map Layers
Control the coloration, level
of geographic detail and
integrated data layers
Data Pane Fields
Geographic fields show in the
data pane with a globe icon
Marks Card Map Options

Adjust map type from the mark drop-down or drag Control options available to end-users on the map
geographic fields to detail to impact the grain of the map related to exploration, scale, and units

Geospatial Use Cases
There are several common use cases that analysts can use to explore and visualize geospatial
data in Tableau, including heat maps, density maps, point distribution charts, and more:
Point Distribution Maps Heatmaps Density Maps
Flow Maps Layered Maps Radius Maps

Standard Maps
There are two types of standard maps available in Tableau: proportional symbol & filled maps
Proportional Symbol Map Filled Map

Symbol maps place symbols at lat/lon coordinates based on Filled maps leverage polygons to produce shaded area maps.
geographic fields used in the view (i.e. Country/State/City/Zip) Most standard geographic and geometric fields can be used to
produce filled maps (except city)

Spatial Functions
Spatial functions can be used for advanced analysis or to combine spatial and non-spatial data
Converts data from latitude and longitude

MAKEPOINT columns into spatial objects
=MAKEPOINT ([Latitude], [Longitude])
MAKELINE Generates a line mark between two points =MAKELINE ([Geometry 1], [Geometry 2])
Returns distance measurement between two

DISTANCE points in a specified unit (m, km, mi, ft)
=DISTANCE ([Geometry 1], [Geometry 2],”mi”)
Returns a circular shape with a radius

BUFFER determined by defined distance and units
=BUFFER (MAKEPOINT ([Latitude], [Longitude]),10,”mi”)

Custom Territories
Custom territories allow users to dynamically select or consolidate geographic areas using
traditional Tableau grouping functionality
Geographic Group
Groups are created using
geographic fields (i.e. State/Zip)
Custom Territory Visualization

Remove other geographic fields except the new Geo
Group (polygons will form around the defined areas)

Map Hierarchies
Mapping hierarchies allow users to drill into custom geographic levels of detail within a map
1 Create Hierarchy 2 Add Geographic Fields 3 Activate Hierarchy

Geospatial Customization
Geospatial customization provides additional mapping options and tool extensibility including
custom formatting, 3rd party maps, and custom background images
Custom Formatting Background Images
• Advanced formatting tools and 3rd party • Custom backgrounds allow you to customize
maps enable a deep level of visuals or create non-traditional maps (i.e.
customization within Tableau Desktop office spaces or factory floor plans)

Map Layers
Map layers can be customized to format map colors, boundaries, labels and built-in data layers
Background
Default mapbox/openstreet
map styles built into Tableau
NOTE: Offline maps are

available in the main Map
menu as well Map Layers
Dimensional layers that can be
Data Layer revealed/hidden by the developer (i.e.
land cover, coastline, zip-code
Additional US layers can be boundaries)
displayed behind the data
set’s metrics and dimensions NOTE: Some layers will not be
visible until users zoom into a map

Background Maps | WMS
Background maps from WMS (“Web Map Services”) allow developers to embed hosted custom
map backgrounds inside Tableau
1 Add WMS Server 2 Layer data & visuals
https://www.terrestris.de/en/openstreetmap-wms/
PRO TIP: You can save WMS connections as .tms files to allow others to easily connect to WMS map connections

Background Maps | Mapbox
Mapbox is a 3rd party tool offering extensive map customization options with a free user account
1 Create Account 2 Customize Map 3 Add Mapbox WMS
Mapbox’s free account allows for 50,000 views / 200,000 tile requests per month.
This is important to consider before integrating with production / client work!

Background Images
Background images are custom visual layers behind your coordinate-based data points
Coordinates
X/Y coordinates can be plotted on top of a
custom background image (using native
Tableau annotations or 3rd party tools
Marks
Unique location dimensions
are added along with desired Native Tableau Annotation:
measure values/metrics
3rd Party Coordinate Generation:

(Interworks Drawing Tool)

Common Mapping Issues
Common mapping issues can occur when data contains unknown or ambiguous spatial points
Edit Locations
Use edit locations to either add spatial fields to help
Tableau identify locations (i.e. adding state to a city map
when duplicate cities exist) or define lat/long points
Filter Data
Filter out the lat/long points in question (useful for true errors
or false records, but be careful not to exclude valid records!
Show data at default position

Null coordinates will default to “Null Island”, a generic
point on Tableau maps that will include missing data
(NOT RECOMMENDED)
PRO TIP: Zip codes are often the source of nulls. Be sure to code zips as strings and/or pad zeros for zips less than 5 characters!

HOMEWORK: Geospatial Mapping
THE As part of your analytics side-hustle, you’ve been hired as an analytics consultant by FEMA
SITUATION (Federal Emergency Management Agency) as part of the National Dam Safety Program (NDSP).
For this project, you will be utilizing the National Inventory of Dams database to provide a
THE
public-facing informational campaign tool to the NDSP.
BRIEF
You’ve been asked to conduct a geospatial analysis to identify high risk dams within
communities targeted for potential rehabilitation grants.

OBJECTIVE • Create spatial joins between data sources
• Leverage distance-based radius filtering & mapping visuals
• Customize mapping visuals using geospatial formatting
• Create custom geo-hierarchies

Advanced Calculations
Advanced calculations go beyond basic Tableau functions to help solve specific, complex use cases
Regular Expressions Table Calculations LOD Expressions
• Regular expressions (RegEx) • Table calcs allow you to • Level of detail (LOD)
leverage text-based search create custom calculations expressions allow users to
pattern algorithms to that exist within the scope dynamically control the
find/replace string values of a specific visual grain of your analysis

Regular Expressions
Regular expressions (RegEx) use special characters to define search patterns and find/replace text
REGEXP_EXTRACT ( [Customer Email] ,’ ( \w + (@adipiscing) + \. [a-z] {2,4} ) ')
RegEx Function Target Field Metacharacters Capture Group Character Class Quantifiers
Tableau regular Tableau field or Denoted by a character Denoted by Denoted by square Defines the acceptable
expression parameter that is preceded by a backslash, parentheses, these are brackets, these can be used number of previous
function used to the subject of the these determine the used to capture exact in place of character literals characters,
call a specific regular expression matching logic: word/text matches: (exact word matches): metacharacters, groups or
action: sets:
REGEXP_REPLACE . Match any character (abc) [abc] any of a, b, or c * match preceding 0 or more times
REGEXP_MATCH \d Match any digit (Douglas) [^abc] not a, b, or c + match preceding 1 or more times
REGEXP_EXTRACT \w Match a word (Apple) [a-g] character between a and g {n} match preceding n times
REGEXP_EXTRACT_NTH \s Match a whitespace [1-3] number between 1 and 3 {n, } match preceding n or more times
\t Match a tab {n,m} match preceding between n and m
\n Match a new line ? Match preceding 0 or 1 times
PRO TIP: Tools like regexr.com or ICU Project Userguide are great resources to help you master regular expression syntax!

Regular Expressions Functions
Regular expressions in Tableau leverage four pre-defined functions, which represent the most
common RegEx use cases: Replace, Match, Extract and Extract_Nth
Returns a copy of the given string where the regular expression pattern is replaced by the
REGEXP_REPLACE replacement string
=REGEXP_REPLACE ( 'abc 123', '\s', '-') = 'abc-123'
Returns true if a substring of the specified string matches the regular expression pattern
REGEXP_MATCH
=REGEXP_MATCH ('-([1234].[The.Market])-','\[\s*(\w*\.)(\w*\s*\])’) = true
Returns only the portion of the string that matches the regular expression pattern
REGEXP_EXTRACT
=REGEXP_EXTRACT ('abc 123', '[a-z]+\s+(\d+)') = '123'
Returns the portion of the string that matches the regular expression pattern. The substring is
REGEXP_EXTRACT_NTH matched to the nth capturing group, where n is the given index.
= REGEXP_EXTRACT_NTH ('abc 123', '([a-z]+)\s+(\d+)', 2) = '123'

REVIEW: Table Calculations
Table calculations are sheet-based computations that only evaluate within a given visualization
• Table calcs affect how values are computed within a visual, but do not change the underlying data
• Table calcs can be created from the Quick Table Calculation or Add Table Calculation menus, or as a formula
Extract Filters Quick Table Calculation: Add Table Calculation: Calculation-Based:
Data Source Filters
Created in the calculation pane

Context Filters and then pulled into the view
Dimension Filters
Measure Filters
Table Calc Filters Quick access to common table calculation Includes more options than
options, based on fields in the view Quick Table Calculations

REVIEW: Addressing & Partitioning
When adding table calculations, you must use all dimensions in the level of detail for either
partitioning (scoping) or for addressing (direction)
Pane / Partition Across then down
Addressing Fields
Define the direction in which a
calculation is evaluated (i.e
down, across, across then down)
Partitioning Fields
Define the level or group at
which table calcs are evaluated
(i.e cells, tables, panes)
PRO TIP: Re-order the addressing fields to

change the direction of the calculation

Position-Based Table Calculation Functions
Position-based table calcs are valuable tools for advanced analytics, such as calculating
metrics like Employee CAGR (continuous average growth rate)
((End Value/Beginning Value)^(1/n)) – 1 LOOKUP

Returns the value of the
given expression in a target
row, specified as a relative
offset from the current row
FIRST
Returns the number of rows
LAST
from the current row to the Returns the number of rows
first row in the partition from the current row to the
last row in the partition

PRO TIP: Index Top Filter
Index is a versatile table calculation which returns the index for the current row in the partition
• One of the most common ways to leverage index is in place of a TOP function
Extract Filters
Data Source Filters
Context Filters
Dimension Filters
Measure Filters
Table Calc Filters

REVIEW: Level of Detail Expressions
LOD expressions allow users to control the granularity at which a calculation is computed
• Unless LOD expressions are used, calculations will always compute at the level of detail shown in the view
Extract Filters Calculation Syntax:
Data Source Filters
Context Filters
Sets, Conditional Filters,
Top N, Fixed LOD
Dimension Filters
Include/Exclude LOD, Level of Detail Element Dimension Declaration Aggregate Expression
Data Blending
Specified LOD types are Defines the grain at which Calculation to be performed
Measure Filters Include, Exclude and Fixed the data is aggregated at the grain defined
Table Calc Filters

Choosing an LOD Type
LOD expressions come in three flavors to define the granularity at which the calculation will be
performed: Fixed, Exclude, and Include
Coarser Grain
Exclude LOD (More Aggregated)
Exclude
Calculated at a coarser grain (higher)
than the view. These calculations are
impacted by filters/dimensions in the view. Repeat
Values
View View level of detail
Include LOD Fixed
Calculated at a finer grain (lower) than
the view. These calculations are impacted
by filters/dimensions in the view. Aggregate
Include
Fixed LOD Finer Grain
(Less Aggregated)
Calculated at a finer or coarser grain than
the view. These calculations are NOT
impacted by filters/dimensions in the view
unless specifically called out or in context!
PRO TIP: Use table-scoped LOD calculations by adding curly brackets around
a field {min([Order Date])}, instead of a fixed LOD {FIXED : min([Order Date])}

LOD Use Case | Benchmarking & Time Flags
Benchmarking using LOD expressions allows users to isolate certain dimensions or measures for
comparative analyses
• Along with benchmarking, time flags can allow for dynamic benchmarked dates to be applied to measures
Benchmarking Time Flags

Use nested IF/LOD expressions to create benchmarks Create static time flags using IF expressions, or use LOD calculations to
against a specific time periods or dimensions allow them to dynamically update as the data evolves
Prior Year Benchmark: Static Flag:
Dynamic Flag:

LOD Use Case | Customer Order Frequency
Customer order frequency leverages LODs to segment customers into buckets for further analysis
or targeting (i.e. marketing, site optimization, promotions, etc.)
FIXED LOD
By fixing on [Customer Id] we can segment order
frequency into buckets to be analyzed as bars
Nested Table Calculation

A pareto-style nested table calc applies a running
total and percent of total to add depth to the
magnitude of the [Orders Per Customer] metric

LOD Use Case | Cohort Analysis
Cohort analysis uses LODs to segment customers based on tenure

• Cohorts can be used to analyze key metrics like customer lifetime value (LTV), retention and churn
FIXED LOD
By fixing on [Customer Id] we can create customer
cohorts and layer them as stacked bar colors

LOD Use Case | New Customer Acquisition
New customer acquisition allows analysts to track net new customers, over time or segmented
by various dimensions (i.e. markets, traffic sources, etc.)
Nested Fixed LOD

By fixing on the customer acquisition date, we isolate
the initial order date and filter on new customers only
Table Calculation
Customer count is placed on rows and layered
with a running total to show cumulative trending

HOMEWORK: Advanced Calculations
THE The Boston Police Department (BPD) commissioner is looking to analyze employee
SITUATION earnings in his department. Concerns about pay increases and comparisons to the
Boston Fire Department have recently raised public interest in this topic.
THE You’ve been asked to analyze a multi-year employee earnings dataset provided by the
BRIEF city of Boston. Since this is public data, the BPD wants to ensure that reports are 100%
accurate and crystal clear in order to promote transparency within the department and
to the general public.

OBJECTIVE • Create regular expressions to find / match text
• Leverage table calculations to create a CAGR analysis
• Use LODs to create cohort, benchmarking, and time flag tools

Predictive Analytics
Predictive analytics is about determining the likelihood of future outcomes based on historical data
• Tableau Desktop provides both internal and external tools to enable predictive analytics
Internal Tools External Tools
• Tableau includes several out-of-the box • Tableau also integrates with third-party
predictive analytics tools, including: programming languages and tools, including:
• Regression (trend lines) • Python
• Forecasting • R
• Control charts • Matlab
• Cluster analysis • etc.

Regression Models
Regression models in Tableau are represented in the form of trend lines

• Regression models are statistical models used to estimate trends/relationships between variables.
Linear Logarithmic Exponential Polynomial Power

Simplest regression model that Curved lines of best fit that Curved lines of best fit most Curved lines of best fit most Curved lines of best fit most
estimates the relationship increase quickly and then level suitable when rate of change suitable when variables suitable when dependent
between two variables off gradually. Avoid logarithmic between variables rises at have a fluctuating variable increases at pre-
(increasing or decreasing). Use trends when values are increasingly higher rates. Avoid relationship (data will determined rate. Avoid power
when data resembles linear negative as they will be exponential trends for negative appear to have a “bounce” trends for negative values
pattern. excluded. values (will be filtered out). in it). (will be filtered out).
Example: If one loaf of bread Example: Efficiency of Example: Computing power of Example: Revenue over Example: The trajectory of a
costs $2, 4 loafs will cost $8 warehouse workers over time microchips over time (power time for a new product skipped stone over time
(eventually maxes out) doubles every two years – launch
Moore’s Law)

Trend Lines
Trend lines reveal the prevailing direction or “line of best fit” for a series of data points
• Trend lines can be especially useful when your data is “noisy”, or when a trend isn’t immediately clear
• NOTE: Trend lines require two continuous variables (if using time, needs to be continuous)
Is my trendline ACCURATE?
R-Squared
R-Squared explains how the variance of one
variable explains the variance of another, ranging
from 0-1 (closer to 1 = more accurate)
Example: If R-Squared = 0.50, then only half of the
observed variation can be explained by the model's
inputs
P-Value
P-Value is the probability of obtaining results as
extreme as the observed results, ranging from 0–1
(closer to 0 = more accurate)
A P-Value of <0.05 is a common standard for
model accuracy and means that your trend-line is
strong predictor for future data points

Forecasting
Forecast algorithms evaluate patterns in historical data to project potential future data points
• Tableau provides several drag-and-drop forecast tools in the analytics pane
Adding a Forecast: Customization:

Forecast Length
Choose how long forecasts will extend; Exactly and
Until will provide fixed dates, and Automatic will
determine a length via forecast results
Source Data
Aggregate by gives you date granularity options
down to the second. You can also ignore a defined
number of data points and fill missing values with
zero
Forecast Model
• Automatic (default)
• Automatic without Seasonality – Removes
Seasonality
• Custom – Adjustable trend and seasonality options:
Exact dates are not supported in Tableau for forecasting. Use truncated dates like Feb 2017 or
continuous dateparts (Year, Year + Quarter, etc.). At least 5 datapoints are required for a forecast, • Additive – Contributions of the model are
at least 2 seasons for seasonality, and at least 24 months for a 12-month seasonal cycle. summed
• Multiplicative – Some contributions are multiplied

Control Charts
Control charts are a statistical tool designed to monitor manufacturing or business processes to
ensure they are in a state of control (within limits)
• Control charts can be generated in Tableau using reference bands
Adding a Reference Band: Customization: Scope

Entire Table, Per Pane, or Per Cell
Value
Percentage, Percentile, Quantile, or Std. Dev
Label
None, Computation, Value, or Custom
Tooltip
Automatic, Custom, or None
Formatting
Line, Fill Color, Fill Above/Below, and
Symmetry
Show Recalculated Band
Recalculate based on the user’s selection

Clustering
Clustering is a method of partitioning observations (marks) into clusters around the nearest
mean (centroid)
• Observations within clusters are more similar to each other than to those in other clusters
Adding a Cluster: Customization:

Variables
The fields Tableau uses to compute clusters
(add/remove variables to adjust clusters)
NOTE: In the Describe Clusters dialog box,

variable value grouping can be reviewed
before modifying the cluster variables
Number of Clusters
The number of clusters will be determined
automatically (using the Calinski-Harabasz
criterion) unless explicitly defined by the user
NOTE: If a categorical value (dimension) has

more than 25 unique values it will be
PRO TIP: Drag the cluster pill from the marks card to the data pane to create a cluster disregarded in the cluster calculation
group that can be used in other sheets. Just refit the data and you’re good to go!

R & Python Integration
R and Python are popular programming languages for performing complex statistical analysis
• R and Python functions and models can be used in Tableau via calculated fields
1 Install Platform: 2 Install Server: 3 Connect to Service: 4 Add Calculated Fields:
install.packages(“Rserve”) SCRIPT_BOOL
library(Rserve)
Rserve() SCRIPT_INT
https://cloud.r-project.org/index.html
SCRIPT_REAL
SCRIPT_STR
https://github.com/tableau/TabPy
https://www.python.org/downloads/
Remember that R/Python are complex languages which require a base-level proficiency to be utilized
inside Tableau. Tableau leverages calculated fields to run R/Python Scripts, but users should know the
languages first to effectively utilize R/Python inside Tableau.

Dashboard Extensions
Dashboard extensions are web applications created by Tableau partners, designed to help users
customize and enhance standard dashboards
1 Download Extension from Gallery: 2 Add Extension to My Tableau Repository:
3 Add to Dashboard & Configure:
Most Tableau Extensions DON’T work with Tableau

Public, and many require some level of external data
sharing or cost for using their services!
https://extensiongallery.tableau.com/extensions?

HOMEWORK: Predictive Analytics
THE The World Health Organization has been generating significant press for a recent study
SITUATION on global life expectancy, and has asked you to help conduct some statistical modeling
and predictive analysis to better understand the data
THE For this project you’ll need to apply predictive analytics tools in Tableau to explore trends,
build forecasts, and conduct a cluster analysis using the WHO life expectancy dataset.
BRIEF
Your job is to find key nuggets of information to help guide the team as they target regions
to investigate in their future research.

OBJECTIVE • Create trends to identify patterns in the data
• Leverage forecasts to project future data points
• Apply clustering to segment WHO data

Maven Advanced Tableau

Uploaded by

Copyright:

Available Formats

Maven Advanced Tableau

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Maven Advanced Tableau

Uploaded by

Copyright:

Available Formats

FOR BUSINESS INTELLIGENCE

With Award-Winning Tableau Expert Dustin Cabral

*Copyright Maven Analytics, LLC

Additional resources include:

Downloadable Ebook to serve as a helpful reference when you’re offline or on the go

*Copyright Maven Analytics, LLC

1 Table Relationships Explore Tableau Relationships and data modelling tools

Create dynamic dashboard views leveraging multiple

Develop custom mapping solutions including buffer

Leverage complex calculations to deliver actionable insights for

Use native and integrated predictive tools in Tableau including

*Copyright Maven Analytics, LLC

THE Use Tableau Desktop to:

*Copyright Maven Analytics, LLC

1 This course is designed to cover advanced topics in Tableau Desktop

3 This course is primarily geared towards data analysis and visualization

4 We will not cover Tableau Server/Online as part of this course

*Copyright Maven Analytics, LLC

*Copyright Maven Analytics, LLC

Logical Layer represents the canvas for

Physical Layer represents the canvas for

The Physical Layer physically joins tables together,

*Copyright Maven Analytics, LLC

*Copyright Maven Analytics, LLC

*Copyright Maven Analytics, LLC

Dirty Data can make relationships more

Matching fields must share the same data

Drag logical tables into the canvas:

Define matching fields

Set Performance Options

*Copyright Maven Analytics, LLC

*Copyright Maven Analytics, LLC

*Copyright Maven Analytics, LLC

Contextual Joins Smart Aggregations

Retain All measure values are retained

Unmatched dimensions can be recovered by using

Unmatched measure values are represented as zero

Unmatched dimension/measure values can be

*Copyright Maven Analytics, LLC

Left Join (“Remain”) Inner Join (“Relevant”)

*Copyright Maven Analytics, LLC

Relationships leveraged in the view

Sales by Store [Sales] Store Lookup [Neighborhood]

*Copyright Maven Analytics, LLC

Relationships leveraged in the view

Store Lookup Customer Lookup

*Copyright Maven Analytics, LLC

Relationships leveraged in the view

Sales by Store [Count()]

Customer Lookup [Home Store]

*Copyright Maven Analytics, LLC

Relationships leveraged in the view

Sales by Store [Count()] Store Lookup [Count()]

Customer Lookup [Home Store]

*Copyright Maven Analytics, LLC

Relationships leveraged in the view

Store Lookup [Store Type]

*Copyright Maven Analytics, LLC

Relationships leveraged in the view

Sales by Store [Count()] Store Lookup [Neighborhood]