The Tableau Workshop

A practical guide to the art of data visualization with Tableau

Sumit Gupta, Sylvester Pinto, Shweta Sankhe-Savale, JC Gillet, Kenneth Michael Cherven

• Editore: Packt Publishing
• Anno: 2022
• ISBN:
  9781800209336
• Edizione: 1
• Argomento: Computer Science
• Categoria: Data Visualisation

Learn how to bring your data to life with this hands-on guide to visual analytics with Tableau

Key Features

• Master the fundamentals of Tableau Desktop and Tableau Prep
• Learn how to explore, analyze, and present data to provide business insights
• Build your experience and confidence with hands-on exercises and activities

Book Description

Learning Tableau has never been easier, thanks to this practical introduction to storytelling with data. The Tableau Workshop breaks down the analytical process into five steps: data preparation, data exploration, data analysis, interactivity, and distribution of dashboards. Each stage is addressed with a clear walkthrough of the key tools and techniques you'll need, as well as engaging real-world examples, meaningful data, and practical exercises to give you valuable hands-on experience.

As you work through the book, you'll learn Tableau step by step, studying how to clean, shape, and combine data, as well as how to choose the most suitable charts for any given scenario. You'll load data from various sources and formats, perform data engineering to create new data that delivers deeper insights, and create interactive dashboards that engage end-users.

All concepts are introduced with clear, simple explanations and demonstrated through realistic example scenarios. You'll simulate real-world data science projects with use cases such as traffic violations, urban populations, coffee store sales, and air travel delays.

By the end of this Tableau book, you'll have the skills and knowledge to confidently present analytical results and make data-driven decisions.

What you will learn

• Become an effective user of Tableau Prep and Tableau Desktop
• Load, combine, and process data for analysis and visualization
• Understand different types of charts and when to use them
• Perform calculations to engineer new data and unlock hidden insights
• Add interactivity to your visualizations to make them more engaging
• Create holistic dashboards that are detailed and user-friendly

Who this book is for

This book is for anyone who wants to get started on visual analytics with Tableau. If you're new to Tableau, this Workshop will get you up and running. If you already have some experience in Tableau, this book will help fill in any gaps, consolidate your understanding, and give you extra practice of key tools.

1. Introduction: Visual Analytics with Tableau

Overview

In this chapter, you will learn about Visual Analytics and why it is important to visualize your data. You will connect to data using Tableau Desktop and familiarize yourself with the Tableau workspace. By the end of this chapter, you will be well acquainted with the Tableau interface and some of the fundamental important concepts that will help you get started with Tableau. The topics that are covered in this chapter will mark the start of your Tableau journey.

Introduction

At a very broad level, the whole data analytics process can be broken down into the following steps: data preparation, data exploration, data analysis, and distribution. This process typically starts with a question or a goal, which is followed by finding and getting the relevant data. Once the relevant data is available, you then need to prepare this data for your exploration and analysis stage. You might have to clean and restructure the data to get it in the right form, maybe combine it with some additional datasets, or enhance the data by creating some calculations. This stage is referred to as the data preparation stage. After this comes the data exploration stage. It is at this stage that you try to see the composition and distribution of your data, compare data, and identify relationships if any exist. This step gives an idea of what kind of analysis can be done with the given dataset.

Typically, people like to explore the data by looking at it in its raw form (that is, at the data preparation stage); however, a quick and easy way to explore the data is to visualize it. Visualizing the data can reveal patterns that were difficult to recognize in the raw data.

The data exploration stage is followed by the data analysis stage, in which you analyze your data and develop insights that can be shared with others. These insights, when visualized, will enable easier interpretation of data, which in turn leads to better decision making. In very simplistic terms, the process of exploring and analyzing the data by visualizing it as charts and graphs is called "visual analytics." As mentioned earlier, the idea behind visualizing your data is to enable faster decision making. Finally, the last step in the data analytics cycle is the distribution stage, wherein you share your work with other stakeholders who can consume this information and act upon it.

In this chapter, we will discuss all these topics in detail, starting with a further exploration of the value of the titular process.

The Importance of Visual Analytics

As mentioned earlier, "Visual Analytics" can be defined as the process of exploring and analyzing data by visualizing it as charts and graphs. This enables end users to quickly consume the information and, in turn, empowers them to make quicker and better decisions.

In this section, you will learn why data visualization is a better tool for evaluation than looking at large volumes of data in numeric format.

All of us have at some point heard the expression "A picture is worth a thousand words." Indeed, it has been found that humans are great at identifying and recognizing patterns and trends in data when consumed as visuals as opposed to large volumes of data in tabular or spreadsheet formats.

To understand the importance and the power of data visualization/visual analytics, let's look at one of the classic examples: Anscombe's Quartet. Anscombe's quartet is comprised of four distinct datasets with nearly identical statistical properties, yet completely different distributions and visualizations.

Note

This was developed in 1973 by an English statistician named Francis John (Frank) Anscombe, after whom it was named.

Let's take a deeper look at these datasets.

Figure 1.1: A screenshot showing the datasets used in Anscombe's quartet

As you can see in the preceding figure, each dataset consists of 11 X and Y points. Now, if you were to analyze these datasets using typical descriptive statistics such as mean, standard deviation, and correlation between X and Y, you would see that the output is identical.

Figure 1.2: A screenshot showing descriptive statistics of the Anscombe's quartet data

Looking at the preceding figure, you can see the following:

• The mean of X for each dataset is 9 (exact accuracy).
• The standard deviation for X for each dataset is 3.32 (exact accuracy).
• The mean of Y for each dataset is 7.50 (accurate up to two decimals).
• The standard deviation for Y for each dataset is 2.03 (accurate up to two decimals).
• The correlation between X and Y for each dataset is 0.816 (accurate up to three decimals).

So, by looking at the above statistical inferences, you would assume that these datasets are identical until you decide to visualize each of them, the results of which are displayed below.

The images show how these datasets appear when visualized as graphs. Now, let's compare each of these visualizations side by side so that you can see how different each of these datasets really are.

Figure 1.3: A screenshot showing a graphical representation of all four datasets of Anscombe's quartet

The preceding example highlights how data visualization can help uncover patterns in data that it was not possible to see by simply looking at the numbers and/or just analyzing the data statistically. This is exactly why Francis Anscombe created his "quartet." He wanted to counter the argument that "numerical calculations are exact, but graphs are rough," which, back then, was a quite common impression among statisticians.

Next, take a look at one more example of how visualizing data helps us find quick insights. Refer to the following figure:

Figure 1.4: A screenshot of a grid view showing the marketing expense and profitability for products across markets

In the preceding figure, you can see a grid view of fields such as Product Type, Product, Market, Marketing, and Profit. In the data that you have used, Marketing is the money that is spent on any marketing efforts to promote products, and Profit is the profit generated after those marketing efforts. Further, these values are broken down by dimensions such as Product Type, Product, and Market. The idea is to evaluate how each product is doing in terms of Marketing and Profit across different markets.

Now, displaying this information in a grid format, as shown above, results in 84 numbers being shown in the view, and doing any kind of comparison across these 84 numbers is going to be very difficult. So, imagine you want to find out whether there are any products in any specific markets where losses are made even after spending significant money on the marketing efforts. Then you will end up comparing these numbers horizontally as well ...