Different Charts for Different Data

This workshop will introduce you to Tableau Software and the wide range of chart types it can create. We’ll focus on which visualizations work best for different types of data, progressing from bar graphs to combination charts.

Start
Here

Before we dive into building visualizations with Tableau, here’s a bit of information about how to use this guide and the building blocks that make up most visualizations you’ll see.

How to use this guide

We move pretty quickly during the workshop. There’s a lot to cover in just 90 minutes! If you miss a step, don’t worry. You can find all the materials we discuss here. Use this guide to solidify your skills or as a reference this year as you dive into your own Tableau projects.

BALTIMORE’S 311 DATA

This workshop uses Baltimore’s 311 data from 2016. Click here to download the Excel spreadsheet we use during the workshop. The original data as well as its documentation can be found here. Thank you to Open Data Baltimore for making this data available to the public.

TABLEAU WORKBOOK

The complete Tableau workbook we use during the workshop can be found here. Just go to the bottom right hand corner and download your own copy. Then you can deconstruct each visualization we built with a fresh version.

STEP-BY-STEP GUIDE

The guide that follows documents the key steps that go into building each of the different chart types we create during the workshop. If you get stuck, download a copy of the original workbook and compare your worksheet to the ones there. Good luck!

WHAT MAKES A VISUALIZATION?

The basic pieces of nearly all data visualizations can be broken down into three kinds of marks: points, lines, and areas. From there, we use position, size, color, and shape to tell multiple marks apart. Check out how these fundamental components are leveraged in these great examples from Tableau Public.

HARRY POTTER SPELLS

Skyler Johnson

HOME RUN RECORDS

Curtis Harris

NEW YORK CITY TREES

Ann Jackson and Josh Jackson

US PRESIDENTIAL ELECTIONS

Jacob Olsufka

ENDANGERED ANIMALS

RJ Andrews

Comparing
Categories

Let’s start with one of the most common data visualization tasks: comparing categories. We’ll dive into the categorical divisions in Baltimore’s 311 data by creating a basic bar chart, 100% stacked bar chart, and a treemap.

Bar Chart

The workhorses in the data viz world, bar charts do a great job comparing the small groups of values. The human eye does a great job of differentiating lengths, so we can glance at a well-sorted bar chart and immediately understand what’s going on. Here are the steps we used to make a bar chart with a filter.

100% Stacked Bar Chart

When the question shifts from how big or small a category is to how a given total splits out by percentages, 100% stacked bar charts become helpful. In our example below, it doesn’t matter how many service requests each agency received, only the different percentages of how they received them.

Let's build a stacked bar that looks at what percentage of each agency's Agency service requests where submitted by phone, mass entry, mobile apps, etc.

Sort the bars using the icon on the x-axis

Change the view from Standard to Entire View

Select the last seven labels and click the paperclip to group them

Right click the new group to change the alias

Replace the concatenated label with simply "Other"

Add the quick table calculation Percent of Total to Number of Records on Columns

Now the bars represent the percentages instead of values

Drag Agency to Rows and set it in front of Methods Received (group)

Drag Methods Received (group) from the rows shelf to Color

Use the dropdown menu to sort the Methods Received

Set up the sort to descend based on the sum of the number of records

Select the Miller Stone palette and click Assign Palette

Click Color again and add white borders between each section

Change the Percent of Total calculation to Compute with Table(across) instead of Table(down)

The color legend will let you highlight one Method at a time

Hold down shift to highlight more than one category at a time

Treemap

When bar charts start to have too many categories to be useful, it’s usually time to try a treemap. More precise than its word cloud and bubble chart cousins, the treemap does a great job of showing how diverse a variable can be without losing track of the heavy hitters.

Use the dropdown menu on Agency to click Sort

Sort the Agencies by the sum of the number of records

Assign the Miller Stone palette to the sorted agencies

Drag Agencies below SR Type on the Marks card to change the treemap's order of operations

Add Number of Records to Tooltip and add the quick table calculation Percent of Total

Add another Number of Records pill to the tooltip (this one won't yet have a delta one it)

Add a Rank quick table calculation to this second Number of Records pill

Use the dropdown menu to edit the table calculation by using both the SR Type and Agency under Specific Dimensions

Now it's time to clean up our tooltips. Here's what they look like by default.

By clicking tooltip and editing the text box, they can look like this

Finally, click the SR Type dropdown menu and select Show Highlighter

Try typing a type of SR Request into the new wildcard text box (e.g., street)

Visualizing
Time

Now let’s move on to visualizing data as it relates to time. Date and time variables are super flexible in Tableau. Just keep an eye on what’s discrete and what’s continuous.

Line Chart

The most classic chart type for longitudinal data, the line chart is a great place to start when it comes to visualizing changes over time. The human eye can detect very subtle shifts in how a line moves up or down.

First take the Created Date to Columns. It will default to a discrete year.

Click the dropdown menu and change the date format to the continuous month.

The Created Date's aggregation will become month and the pill will turn green because it's continuous data.

Now take Number of Records to rows. Tableau will default to a line.

Now add Agency to Color. This chart does a great job showing how the DOT and Health Department have the same number of requests in the summer.

But the smaller departments are hard to distinguish along the bottom. Overplotting like this is the most common problem with line charts, and we'll fix it in our next visualization.

Area Chart

Like line charts, area charts do a great job of displaying changes over time. We can add color to distinguish between different segments in a single area chart. Or, as we’ll do below, we can split out those segments into their own area charts using a tactic called small multiples.

Start by right clicking your line chart and select Duplicate.

Then change the mark type from the automatic line to the area option.

Drag Agency to the Rows shelf and set it before the Number of Records pill.

Now let's adjust the Y-Axis so that we can easily see the shifts over time for each Agency. Right now all the Y axes are identical, which mutes the shifts for the smaller agencies.

Change the Range setting to Automatic to Independent and delete the "Number of Records" text in the Title.

Scroll through the axes to see how they've adjusted to fit each Agency.

Now let's change the granularity of the changes over time from months to week number.

This shift allows us to see specifically intense weeks.

Change the view from Standard to Fit Width to see the full year at once.

When you scroll down to the smaller departments, their shifts throughout the year are now easy to see.

Heat Map

The human eye sees color before it absorbs anything else. A heat map takes advantage of this by filling a typical grid with color instead of plain numbers or text. We’ll still offer our end user all that tabular information in the tooltip. At a glance, we can spot the most intense 311 day in 2016: January 25.

First drag the Created Date to columns. Just like before, it will default to a discrete year.

Change the date format to a discrete Month.

Add another Created Date to Rows and change its formate to a discrete day.

These two variables will make a grid for the entire year. Change the view from Standard to Entire View.

Now add Number of Records to color. There's our heat map.

Click the dropdown menu on Agency and select Show Filter.

Change the filter format to radio button.s

Toggle through the different agencies. You'll see that they have different patterns. For example, housing service requests increase during the summer months, but still retain the weekend lulls we see throughout.

Combination Chart

Last but not least we’ll create a chart that combines a bar chart with a circle marker. Combination charts like this can be especially helpful when there are two variables that need to be considered at the same time. To make this chart, we will also create our first calculated field to determine the number of days between the service requests created date and due date.

Here's what we're trying to build: the blue bars look at the most common service requests for the DOT on 1/25/16 and the yellow dots indicate how many days the department has to complete them.

Start by dragging the Created Date to the filter card and choose the specific Month > Day > Year format.

Drag the Agency pill to the filters card as well and select the Department of Transportation.

Now let's build the first part of the visualization. Drag the SR Type to Rows.

And sort the bars from greatest to smallest using the icon on the x-axis.

Find the up and down arrows between the rows to expand the spacing a bit.

Now for the dots! Right click in a blank section of the data pane and select Create Calculated Field...

Here's the calculation we want. We're figuring out the number of days between the Created Date and the Due Date for each service record (every row).

Add your new variable to the columns shelf and change the aggregation from Sum to Median.

Click the light grey null indicator in the bottom left corner and filter out the data that doesn't have enough information to calculate the difference (usually there's no due date).

If you sort the bars from the fewest days to the most. This new calculation can help us see the department's priorities.

Now let's combine them. Click the datediff dropdown menu and choose Dual Axis.

Click your middle marks card and change the mark type from a dot to a bar.

Now navigate to the bottom marks card for the Date Diff calculation and and make the circle a bit smaller using the size button.

Now navigate to the top marks card and click Edit Colors.

Last but not least, navigate to the Millers palette and select each individual variable and then the color you want to assign to it.

Day
TWO

The second workshop dives into prepping a data set for Tableau. Using just ten records, we’ll work through the most common problems you’ll encounter while Structuring Data for Tableau.

OUR MISSION

To help democratize quantitative knowledge by creating educational resources and experiences that make data visualization–and by extension, data itself–approachable, exciting, and meaningful.

DATA DOZEN