In the professional world, turning massive amounts of data

into something that can be seen and understood is vitally

important. This is where Tableau steps in. It has emerged
as a clear leader in data visualization because it translates
your actions into a database query and expresses the
response graphically.
It also has the unique ability to do ad hoc analysis of millions
of rows of data in just a matter of seconds with Tableau's
Data Engine. Tableau is a rare software platform that is
intuitive and even fun to use, which also enables you to dive
deep into answering complex questions about your data.
Starting with creating your first dashboard in Tableau 9.0,
this book will let you in on some useful tips and tricks, teach
you to tell data stories using dashboards, and teach you how
to share these data stories. Practical examples along with
detailed explanations of how and why various techniques
work will help you learn and master Tableau quickly.

If you want to understand your data using data visualization

and don't know where to start, then this is the book for you.
Whether you are a beginner or have years of experience,
this book will help you to quickly acquire the skills and
techniques used to discover, analyze, and communicate
data visually. Some familiarity with databases and data
structures is helpful, but not required.

Explore and analyze your data by creating

basic and advanced data visualizations
Fix data problems, enhance your
analysis, and create rich interactivity
using custom calculations
Perform effective analysis by joining and
blending data from different sources
Enhance your visualizations with custom
formatting, labels, and annotations

Create meaningful dashboards in Tableau

Extend the value and functionality of
your data
Share your data story using story points and
fully interactive dashboards
$ 49.99 US
32.99 UK

"Community
Experience
Distilled"

C o m m u n i t y

Explore advanced topics such as

sheet swapping, custom maps, and
LOD calculations

Joshua N. Milligan

Who this book is written for

What you will learn from this book

Learning Tableau

Learning Tableau

D i s t i l l e d

Learning Tableau
Leverage the power of Tableau 9.0 to design rich data
visualizations and build fully interactive dashboards

Prices do not include

local sales tax or VAT
where applicable

Visit www.PacktPub.com for books, eBooks,

code, downloads, and PacktLib.

E x p e r i e n c e

Joshua N. Milligan

In this package, you will find:

The author biography

A preview chapter from the book, Chapter 8 'Adding Value to Analysis Trends,
Distributions, and Forecasting'
A synopsis of the books content
More information on Learning Tableau

About the Author

Joshua N. Milligan has been a consultant with Teknion Data Solutions since 2004,
where he currently serves as a team leader and project manager. With a strong
background in software development and custom .NET solutions, he uses a blend
of analytical and creative thinking in BI solutions, data visualization, and data
storytelling. His years of consulting have given him hands-on experience in all aspects
of the BI development cycle, including data modeling, ETL, enterprise deployment,
data visualization, and dashboard design. He has worked with clients in numerous
industries, including financial, healthcare, marketing, and government.
In 2014, Joshua was named a Tableau Zen Master, the highest recognition of
excellence from Tableau Software. As a Tableau-accredited trainer, mentor, and leader
in the online Tableau community, he is passionate about helping others gain insights into
their data. He was a technical reviewer of Tableau Data Visualization Cookbook, Packt
Publishing, and is currently reviewing Creating Data Stories with Tableau Public, Packt
Publishing. His work has appeared multiple times on Tableau Public's Viz of the Day
and Tableau's website. Joshua also shares frequent Tableau tips, tricks, and advice on
his blog at
.

Learning Tableau
The Tableau community is full of individuals passionate about the software. We use
software every dayweb browsers, word processors, e-mail applications, instant
messaging, and numerous other apps. What is it about Tableau that inspires people to
write books and blogs and spend hours volunteering to help others visualize their data?
Tableau is unique in several ways. It is easy and transparent. You can immediately
connect to nearly any data source and start asking and answering questions about your
data in a visual way. It's also intuitive. Its interface allows hands-on interaction with
data, it's easy to get into a flow, and every action uncovers new insights. It's fun! It
allows creativity and gives freedom. You're not locked into chart types and wizards
that give only one path to a solution. Tableau designers feel like artists, with data as
paint and Tableau as a blank canvas.
At the same time, Tableau introduces a paradigm vastly different from traditional BI
tools. This book presents the fundamentals for understanding and working within that
paradigm. It will equip you with the foundational concepts that will help you use
Tableau to explore, analyze, visualize, and share the stories contained in your data.

What This Book Covers

Chapter 1, Creating Your First Visualizations and Dashboard, introduces the basic
concepts of data visualization and multiple examples of individual visualizations,
which are ultimately put together in an interactive dashboard.
Chapter 2, Working with Data in Tableau, shows that Tableau has a very distinctive
paradigm for working with data. This chapter explores that paradigm and gives
examples of connecting to and working with various data sources.
Chapter 3, Moving from Foundational to Advanced Visualizations, expands upon the
basic concepts of data visualization to show you how standard visualization types can
be extended.
Chapter 4, Using Row-level and Aggregate Calculations, introduces the concepts of
calculated fields and the practical use of calculations, and walks through the
foundational concepts for creating row-level and aggregate calculations.
Chapter 5, Table Calculations, proves that table calculations are one of the most
complex and powerful features in Tableau. This chapter breaks down the basics of
scope, direction, partitioning, and addressing to help you understand and use these
to solve practical problems.

Chapter 6, Formatting a Visualization to Look Great and Work Well, shows how
formatting can make a standard visualization look great, have appeal, and communicate
well. This chapter introduces and explains the concept of formatting in Tableau.
Chapter 7, Telling a Data Story with Dashboards, dives into the details of building
dashboards and telling stories with data. It covers the types of dashboards, objectives of
dashboards, and concepts such as actions and filters. All of this is done in the context of
practical examples.
Chapter 8, Adding Value to Analysis Trends, Distributions, and Forecasting, explores
the analytical capabilities of Tableau and demonstrates how to use trend lines,
distributions, and forecasting to dive deeper into the analysis of your data.
Chapter 9, Making Data Work for You, explains that data in the real world isn't always
structured well. This chapter examines the structures that work best and the techniques
that can be used to address data that can't be fixed.
Chapter 10, Advanced Techniques, Tips, and Tricks, builds upon the concepts covered
in the previous chapters. This chapter expands your horizons by introducing numerous
advanced techniques while giving practical advice and tips.
Chapter 11, Sharing Your Data Story, throws light on the fact that that, once you've built
your visualizations and dashboards, you'll want to share them. This chapter explores
numerous ways of sharing your stories with others.

Adding Value to Analysis

Trends, Distributions,
and Forecasting
Sometimes, quick data visualization needs a little deeper analysis. For example, a
simple scatterplot can reveal outliers and clusters of values. However, often, you
want to understand the distribution. A simple time series helps you see the rise
and fall of a measure over time. But many times, you want to see the trend or make
predictions of future values.
Tableau enables you to quickly enhance your data visualizations with statistical
analysis. Built-in features, such as trend models, distributions, and forecasting, allow
you to quickly add value to your visual analysis. Additionally, Tableau integrates
with R, an extensive statistical platform that opens up endless options for the
statistical analysis of your data. This chapter will cover built-in statistical models
and analysis.
This chapter will cover the following topics:

Trending

Forecasting

Distributions

We'll take a look at these concepts in the context of a few examples using some
sample datasets. You can follow and reproduce these examples using this
chapter's workbook.

[ 221 ]

Adding Value to Analysis Trends, Distributions, and Forecasting

Trends
Let's say you are analyzing populations of various countries using the World
Population dataset in the provided workbook. This dataset produces one record
containing the population for each country for each year from 1960 to 2013. Create
a view similar to the one shown in the following screenshot, which shows you the
change in population over time for Afghanistan and Australia. You'll notice that
Country Name has been filtered and added to the Color and Label shelves.

From this visualization alone, you can make several interesting observations. The
growth of the two countries' populations was fairly similar up to 1980. At that point,
the population of Afghanistan went into decline until 1988 when the population
of Afghanistan started to recover. At some point around 1996, the population of
Afghanistan exceeded that of Australia. The gap has grown wider ever since.

[ 222 ]

Chapter 8

While we have a sense of the two trends, they become even more obvious when we
see them. Tableau offers several ways to add trend lines:

From the menu, navigate to Analysis | Trend Lines | Show Trend Lines

Right-click on an empty area in the pane of the view and select Show
Trend Lines

Switch to the Analytics tab in the left-hand side pane and drag and drop
Trend Line on the trend model of your choice (we'll use Linear for now and
discuss the others later in this chapter)

Once you have added the trend line, your view should look like this:

[ 223 ]

Adding Value to Analysis Trends, Distributions, and Forecasting

Trends are calculated by Tableau after querying the data source. Trend lines are
drawn based on various elements in the view:

The two fields that define x and y coordinates: The last (right-most) field
on Rows and Columns will define the axes that give Tableau x and y
coordinates to calculate various trend models. In order to show trend lines,
you must use a continuous (green) field or discrete (blue) date fields and
have one such field on both Rows and Columns. If you use a discrete (blue)
date field to define headers, the other field must be continuous (green).

Additional fields that create multiple, distinct trend lines: Discrete (blue)
fields on the Rows, Columns, or Color shelves can be used as factors to split
a single trend line into multiple, distinct trend lines.

The trend model selected: We'll examine the differences in models in the
next section.

Observe in the view that there are two trend lines. As Country Name is a discrete
(blue) field on Color, it defines a trend line per color by default.
Earlier, we observed that the population of Afghanistan increased and decreased
within various historical periods. Notice that the trend lines are calculated along the
entire date range. What if we want to see different trend lines for these time periods?
We can force Tableau to draw distinct trend lines using a discrete field on Rows,
Columns, or Color.
Go ahead and create a calculated field called Period that defines discrete values for
the different historical periods and using code like this:
IF Year([Year]) <= 1979
THEN "1960 to 1979"
ELSEIF Year([Year]) <= 1988
THEN "1980 to 1988"
ELSE "1988 to 2013"
END

[ 224 ]

Chapter 8

When you place Period on columns, you'll get a header for each time period, which
breaks the lines and causes separate trends to be shown for each time period. You'll
also observe that Tableau keeps the full date range in the axis for each period. You can
set an independent range by right-clicking on one of the date axes, selecting Edit Axis,
and then checking the option for Independent axis range for each row or column.

In this view, transparency has been applied to Color and the trend lines have been
formatted to make them stand out. Additionally, the axis for Year was hidden (by
unchecking the Show Header option on the field). Now, you can clearly see the
difference in trends for different periods of time. Australia's trends only change
slightly in each period. Afghanistan's trends were quite different.

[ 225 ]

Adding Value to Analysis Trends, Distributions, and Forecasting

Customizing trend lines

Let's take a look at another example that will allow us to consider various options for
trend lines. Create a new sheet and use the Real Estate data source connection to
create a view similar to this one:

Here, we created a scatterplot with the sum of Area on Columns to define the x axis
and the sum of Price on Rows to define the y axis. Address has been added to the
level of detail on the Marks card to define the level of aggregation. So, each mark on
the scatterplot is a distinct address at a location defined by the area and price. Type
has been added to Color. We've also shown the trend lines and are getting one trend
line per color by default. Assuming a good model, the trend lines demonstrate how
much and how quickly Price is expected to rise with an increase in Area.

[ 226 ]

Chapter 8

In this dataset, we have two fields, Address and ID, either of

which define a unique record. Adding one of these fields to the
level of detail effectively disaggregates the data and allows us
to plot a mark for each address. Sometimes, you may not have a
field in the data that defines uniqueness. In these cases, you can
disaggregate the data by unchecking Aggregate Measures from
the Analysis menu.
Alternately, you can use the drop-down menu on each of the
measure fields on Rows and Columns to change them from
measures to dimensions while keeping them continuous. As
dimensions, each individual value will define a mark. Keeping
them continuous will retain the axes required for trend lines.

Let's consider some of the options available for trend lines. You can edit trend lines
by using the menu and navigating to Analysis | Trend Lines | Edit Trend Lines
or clicking/right-clicking on a trend line and then selecting Edit. When you do this,
you'll see a dialog box similar to this:

Here, you have options to select a model type, including applicable fields as
factors in the model, allowing discrete colors to define distinct trend lines, showing
confidence bands, and forcing the y intercept to zero. Experiment with the options
for a bit. Notice how either removing the Type field as a factor or unchecking the
Allow a trend line per color option results in a single trend line.
[ 227 ]

Adding Value to Analysis Trends, Distributions, and Forecasting

You can also see the result of excluding a field as a factor in the following view,
where Type has been added to Rows:

As represented in the left portion of the preceding screenshot, Type is included

as a factor. This results in a distinct trend line for each type of sale. When Type is
excluded as a factor of the same trend line, which is the overall trend for all types,
a trend line is drawn three times. This technique can be quite useful to compare
subsets of data with the overall trend.

[ 228 ]

Chapter 8

Trend models
We'll return to the original view and stick with a single trend line as we consider the
trend models available. The following models can be selected from the Trend Line
Options window:

Linear: We'll use this model if we assume that as Area increases, Price will
increase at a constant rate. No matter how high Area increases, we'll expect
Price to increase such that new data points fall close to the straight line.

Logarithmic: We'll use this model if we believe that there is a "law of

diminishing returns" in effect. That is, area can only increase to a certain
extent before buyers stop paying much more:

[ 229 ]

Adding Value to Analysis Trends, Distributions, and Forecasting

Exponential: We'll use this model to test the idea that each additional
increase in area results in a dramatic (exponential) increase in price:

Polynomial: We'll use this model if we feel the relationship between Area
and Price is complex and follows more of an S-shaped curve, where, though
initially increasing the area dramatically increases the price, at some point the
price levels. You can set the degree of the polynomial model anywhere from 2
to 8. The trend line shown here is a third-degree polynomial:

[ 230 ]

Chapter 8

Analyzing trend models

It can be useful to observe trend lines, but often, we'll want to understand whether
the trend model we've selected is statistically meaningful. Fortunately, Tableau gives
us some visibility into trend models and calculations.
Simply hovering over a single trend line will reveal the calculation as well as P-value
for that trend line.

A p-value is a statistical concept that describes the probability

that the results of assuming no relationship between values
(random chance) are at least as close as results predicted by
the trend model. A p-value of 5 percent (.05) will indicate a 5
percent random chance describing the relationship between
values as well as the trend model. This is why p-values of 5
percent or less are considered to indicate a significant trend
model. If your p-value is higher than 5 percent, then you should
not consider that trend to significantly describe any correlation.

[ 231 ]

Adding Value to Analysis Trends, Distributions, and Forecasting

Additionally, you can see a much more detailed description of the trend model by
navigating to Analysis | Trend Lines | Describe Trend Model from the menu
or using the similar menu from a right-click on the view's pane. When you view the
trend model, you will see the Describe Trend Model window:

You can also get a trend model description in the worksheet

description, which is available from the Worksheet menu or by
pressing Ctrl + E. The worksheet description includes quite a bit
of other useful summary information about the current view.

[ 232 ]

Chapter 8

The wealth of statistical information shown in the window includes a description of

the trend model, the formula, the number of observations, and the p-value for the
model as a whole as well as for each trend line. Note that in the window shown in
the preceding screenshot, the Type field was included as a factor that defined three
trend lines. At times, you may observe that the model as a whole is statistically
significant even though one or more trend lines may not be.
Additional summary statistical information can be displayed
in Tableau Desktop for a given view by showing the summary.
From the menu, navigate to Worksheet | Show Summary. The
information displayed in the summary can be expanded using
the drop-down menu on the Summary card.

Tableau also gives you the ability to export data, including data related to trend
models. This allows you to more deeply, and even visually, analyze the trend model
itself. Let's analyze the third-degree polynomial trend line of the real estate price and
area scatterplot without any factors. To export data related to the current view, use
the menu and navigate to Worksheet | Export | Data. The data will be exported as
a Microsoft Access Database (.mdb) and you will be prompted as to where to save
the file.

[ 233 ]

Adding Value to Analysis Trends, Distributions, and Forecasting

On the Export Data to Access screen, specify an access table name and select
whether you wish to export data from the entire view or the current selection. You
may also specify that Tableau should connect to the data. This will generate the data
connection and make it available with the specified name in the current workbook.

The new data source connection will contain all the fields that were present in the
original view as well as additional fields related to the trend model. This allows us
to build a view such as the following using residuals and predictions:

[ 234 ]

Chapter 8

A scatterplot of predictions and residuals allows you visually see how far each mark
was from the location predicted by the trend line. It also allows you to see whether
residuals are distributed evenly on either side of a zero. An uneven distribution
would indicate problems with the trend model.
You can include this new view along with the original one in a dashboard to
explore the trend model visually. Use the highlight button on the toolbar to
highlight the Address field.

With the highlight action defined, selecting marks in one view will allow you to see
them in the other. You could extend this technique to export multiple trend models
and dashboards to evaluate several trend models at the same time, as shown in the
following screenshot:

You can achieve even more sophisticated statistical analysis,

leveraging Tableau's ability to integrate with R. R is an open source
statistical analysis platform and a programming language with which
you can define advanced statistical models. R functions can be called
from Tableau using special table calculations (all of which start with
SCRIPT_). These functions allow you to pass expressions and values
to a running R Server that will evaluate the expressions using built-in
libraries or custom-written R scripts and return results to Tableau.
You can learn more about Tableau and R integration from this
whitepaper (you will need to register a free account first):
http://www.tableausoftware.com/learn/whitepapers/
using-r-and-tableau

[ 235 ]

Adding Value to Analysis Trends, Distributions, and Forecasting

Distributions
Analyzing distributions can be quite useful. We've already seen that certain table
calculations are available to determine statistical information such as averages,
percentiles, and standard deviations. Tableau also makes it easy to quickly visualize
various distributions including confidence intervals, percentages, percentiles,
quantiles, and standard deviations.
You may add any of these visual analytic features using the Analytics tab
(alternately, you can right-click on an axis and select Add Reference Line). Just like
reference lines and bands, distribution analytics can be applied within the scope of a
table, pane, or cell. When you drag and drop the desired visual analytic, you'll have
options to select the scope and the axis. In the following example, we've dragged and
dropped Distribution Band from the Analytics tab onto the scope of Pane for the
axis defined by Sum(Price):

Once you have selected the scope and axis, you will be given options to change
settings. You may also edit lines, bands, distributions, and box plots by right-clicking
on the analytic feature in the view or by right-clicking on the axis. Here, we'll define
settings for one and two standard deviations above and below the mean:

[ 236 ]

Chapter 8

Each specific Distribution option specified in the Value dropdown under Computation
has unique settings. Confidence Interval, for example, allows you to specify a percent
value for the interval. Standard Deviation allows you to enter a comma-delimited list
of values that describe how many standard deviations are used, and at what intervals.
This, for example, is the result of specifying standard deviations of -2, -1, 1, 2:

[ 237 ]

Adding Value to Analysis Trends, Distributions, and Forecasting

Each axis can support multiple distributions, reference lines, and bands. Here,
first and second standard deviations on both sides of the average (the solid line)
are shown. You'll notice that the Type field defines three panes and the standard
deviations have been set to be calculated per pane.
On a scatterplot, using a distribution for each axis can yield a
very useful way to analyze outliers. Showing a single standard
deviation for both Area and Price allows you to easily see
properties that fall within norms for both, one, or neither.

Forecasting
As we've seen, trend models make predictions. Given a good model, you expect
additional data to follow the trend. When the trend is over time, you can get an
idea about where future values may fall. However, predicting future values often
requires a different type of model. Factors such as seasonality can make a difference
not predicted by a trend alone. Starting with version 8.0, Tableau includes built-in
forecasting models that can be used to predict and visualize future values.
To use forecasting, you'll need a view that includes a date field or enough date parts
for Tableau to reconstruct a date (for example, a Year and a Month field). You may
drag and drop Forecast from the Analytics tab, navigate to Analysis | Forecast |
Show Forecast from the menu, or right-click on the view's pane and select the option
from the context menu.

[ 238 ]

Chapter 8

Here, for example, is the view of the population growth of Afghanistan and Australia
with forecasts shown over time:

Note that when you show the forecast, Tableau adds a forecast icon to the
SUM(Population) field on Rows to indicate that the measure is being forecast.
Additionally, Tableau adds a new special Forecast Indicator field to Color so that
forecast values are differentiated from actual values in the view.
You can move the Forecast Indicator field or even copy it (hold
Ctrl while dragging and dropping) to other shelves to further
customize your view.

[ 239 ]

Adding Value to Analysis Trends, Distributions, and Forecasting

When you edit the forecast by navigating to Analysis | Forecast | Forecast

Options from the menu or when you use the right-click context menu on the view,
you will be presented with various options to customize the trend model, like this:

Here, you have options to set the length of the forecast, determine aggregations,
customize the model, and set whether you wish to show prediction intervals.
The forecast length is set to Auto by default, but you can extend the forecast by
a custom value.
The options under Source Data allow you to optionally specify a different grain of
data for the model. For example, your view might show a measure by year but you
could allow Tableau to query the source data to retrieve values by month and use a
finer grain to potentially achieve better results.
[ 240 ]

Chapter 8

Tableau's ability to separately query the data source to obtain

data at a finer grain for more precise results works well with
relational data sources. However, OLAP data sources are not
compatible with this approach, which is one reason forecasting
is not available when working with cubes.

By default, the last value is excluded from the model. This is useful when you are
working with data where the most recent time period is incomplete. For example,
when records are added daily, the last (current) month is not complete until the final
records are added on the last day of the month. Prior to this last day, the incomplete
time period might skew the model unless it is ignored.
The model itself can be set to Automatic with or without seasonality or can be
customized to set options for seasonality and trend. To understand the options,
consider the following view of Sales by MONTH from the Superstore sample data:

[ 241 ]

Adding Value to Analysis Trends, Distributions, and Forecasting

The data displays a distinct cyclical or seasonal pattern. This is very typical for retail
sales. The following are the results of selecting various custom options:

Much like trends, forecast models and summary information can be accessed using
the menu. Navigating to Analysis | Forecast | Describe Forecast will display a
window with tabs for both the summary and details concerning the model.

[ 242 ]

Chapter 8

Clicking on the link at the bottom of the window will give you much more
information on the forecast models used in Tableau.
Forecast models are only enabled with a certain set of conditions.
If the option is disabled, ensure that you are connected to
a relational database and not OLAP, are not using table
calculations, and have at least five data points.

Summary
Tableau provides an extensive set of features to add value to your analysis. Trend
lines allow you to more precisely identify outliers, determine which values fall
within the predictions of certain models, and even make predictions of where
measurements are expected. Tableau gives you extensive visibility into the trend
models and even allows you to export data containing trend model predictions and
residuals. Distributions are useful to understand how measurements are distributed.
Forecasting allows a complex model of trends and seasonality to predict future
results. Having a good understanding of these tools will give you the ability to
clarify and validate your initial visual analyses.
Next, we'll turn our attention back to the data. We considered very early on how to
connect to data, and we've been working with data ever since. However, we've spent
most of our time working with clean, well-structured data. In the next chapter, we'll
consider how to deal with messy data.

[ 243 ]

Get more information Learning Tableau

Where to buy this book

You can buy Learning Tableau from the Packt Publishing website.
Alternatively, you can buy the book from Amazon, BN.com, Computer Manuals and most internet
book retailers.
Click here for ordering and shipping details.

www.PacktPub.com

Stay Connected: