Tableau 10 Business Intelligence Cookbook: Create powerful, effective visualizations with Tableau 10

To follow this recipe, open B05527_01 – STARTER.twbx file. Use the worksheet called Bar, and connect to the HollywoodMostProfitableStories data source.

The following are the steps to create the top 15 movies bar chart:

Bar charts are probably one of the most common, if not the most common, type of chart that is used to visualize data. Bar charts are best used when you are comparing numeric data that is clearly split between different items or categories. These charts are an easy favorite among the chart types because it allows us to easily look at the edges, and see approximately how much longer one bar is than another, even if the difference is very slight. Bars are also often sorted to make it easy to spot top or bottom items.

Bar charts can be presented either vertically or horizontally. The decision sometimes depends on the number of items being represented. If you have many items, labels can become less readable with vertical bar charts. In Tableau, there is an easy way to switch between vertical and horizontal bar charts. There is a swap icon in the Tableau toolbar that swaps the pills from Rows to Columns and vice versa.

In this recipe, we simply dragged the Film dimension to Columns and the Avg. Rotten Tomatoes % measure to Rows, and Tableau already knew what to do. When you check out your Marks card, you will see that the selection is still set to Automatic, but the actual icon beside it is bars:

To help the reader understand the bar chart better, we can also sort the bars in either ascending or descending fashion. A quick way to sort items in Tableau is by hovering over a column header or an axis, until you see a sort icon appear. In an axis, the first click on the icon sorts the marks in descending fashion. The second click sorts ascending, and the third click sorts it back to data source order. With a column header, which is produced by discrete fields, the fields will be sorted alphabetically in ascending order on the first click, descending on the second click, and then back to data source order on the third click.

In our recipe, we also picked the top 15 highest rated films from our data source. We are setting the expectation that as the reader sees the movies from top to bottom, the ratings also are lower. Tableau makes it easy to pick the top n items on some measure. You can simply right-click on the dimension you want to filter, and select Filter… to open the filter window for more options. Tableau provides a tab called Top, which allows you to keep only the top or bottom n items, based on a measure you specify.

One enhancement we can do with bar charts that Tableau allows us to do very easily is to overlay another measure. In this recipe, we used color to show the Avg Audience Score %. You will notice that although the length of the bars are in descending fashion, the color density doesn't follow the same pattern. This indicates that the audience may not necessarily agree with the Rotten Tomatoes % ratings.

Bar charts can also be created using Tableau's Show Me tool, which by default appears in the top-right corner of your Tableau design area. You simply need to select the fields from your side bar that you wish to visualize, and the possible charts from this combination are enabled in the Show Me tool. The Show Me tool is introduced and discussed in more detail in Appendix C, Working with Tableau 10.

To follow this recipe, open B05527_01 – STARTER.twbx. Use the worksheet called Stacked Bar, and connect to the HollywoodMostProfitableStories data source.

The following are the steps to create the stacked bar chart:

A stacked bar chart is simply a bar chart that has bars that are further segmented or split based on some criteria. Stacked bar charts have related values stacked on top of each other, and are best used when we want to see cumulative totals instead of individual values.

To create a stacked bar chart, you simply need to drop a discrete field onto a formatting card in your Marks card.

The Marks card contains additional properties, and depending on the type of mark, there may be a sixth property that appears. To create segments in a bar, we can add the discrete field in any of the following: Color, Size, Label, and Detail.

The Color tool is often used in stacked bar charts as this splits the bar into different, discernible colors. In this recipe, we started by using a bar chart that represents the top grossing films in 2007-2011. We placed the dimension Genre onto Color, which created the color partitions in our graph based on Genre.

However, depending on what you're displaying, you may also consider using Size:

You can also segment the bars by Label:

You could also use the combination of Color, Size, and Label:

The Detail property will also work, but this does not make the segmentation obvious. It simply creates the very thin borders around the segment—which can be easily overlooked.

I encourage you to be adventurous—see what's possible! Tableau allows you to explore possibilities with ease, which is one thing I really love about the tool. I may not necessarily use that combination, but seeing the combination might open up possibilities for me, or see data in a different light, or it might help uncover insights that were previously buried when we were analyzing using a single formatting card.

It might be tricky to use Show Me to generate a stacked bar chart because the fields placed in Rows, Columns, and Headers are automatically assigned. After the initial chart is generated from Show Me, you may need to swap axes, or move pills around.

In addition, be careful when using stacked bar charts. Stacked bar charts are great when you still want to see that cumulative number. However if you wanted to see and compare each individual segment, it will make your analysis hard because the not all of the segments have the same starting point.

In the previous screenshot, if you highlight just Comedy, you will see that the bars in each year don't start at the same point, therefore it is harder to compare them. For example, can we easily tell that the Worldwide Gross Amount for Comedy is greater for 2010 compared to 2009? It is hard to tell from this graph.

To follow this recipe, open B05527_01 – STARTER.twbx. Use the worksheet called Line, and connect to the MasterTable (FlooddataMasterListrev) data source.

The following are the steps to create a time series graph in Tableau:

By default, Tableau creates a line chart—also called a time series chart—when a Date (or Date & Time) field is placed in either the Rows or Columns shelf. A Date field is presented in the side bar with a small calendar icon, while a Date & Time field will be a small calendar icon with a clock. Line charts are best used when you want to see patterns over time.

In this recipe, when we dragged the Began field onto Rows, the field showed up as YEAR(Began) instead of just the Began field.

Date fields in Tableau have a natural hierarchy—meaning you can roll up to a higher date level like YEAR, or drill down to a lower level like DAY. In this case, we see a blue YEAR(Began) pill, which indicates Tableau presented the years as a series of discrete headers.

When the year field was rendered, the first column value that appeared was Null. A Null means there is no value, that is, this data is missing for some records. One way to remove it is to right-click the column header and select Exclude. Exclude creates a filter, but it is a negation filter. Whatever is checked will be excluded. When you investigate the settings for this filter, you will find that Null is checked, but the bottom checkbox for Exclude is checked.

In this recipe, we also used SUM(Dead) to be presented both in Color and Size. The darker the color and the thicker the line indicates the more fatalities. Adding this visual cue in the line graph allows us to easily identify when the most fatalities occurred, without having to hover over each point in the graph and look at the details.

Tableau is a great visualization tool that gives us a lot of flexibility over how we want to display our graphs. However, it can also be a double-edged sword. Nothing stops us from creating all kinds of chart just because we can. However, we shouldn't. Just because we can does not mean we should. We should identify first our data points, and evaluate the best (or most effective) way to represent them.

Take, for example, the following line chart. The line chart has a vertical sales axis. At the bottom of the graph, we can find the discrete categories.

Is the line chart the best way to represent this information? No. In fact, this is incorrect and grossly misleading. When your audience first looks at this graph, the initial perception is that we are showing the fluctuations of sales over time. There is also a notion of relatedness, that is, that the category headers are somehow related. This isn't true. Categories are nominal. There is no inherent order in these classifications. When we move the category labels around, it shouldn't affect the message that is being relayed by the graph. This particular graph can be best represented as a bar chart.

To follow this recipe, open B05527_01 – STARTER.twbx. Use the worksheet called Scatter Plot, and connect to the Top Athlete Salaries (Global Sport Finances) data source.

The following are the steps to create the scatter plot presented in this recipe:

Scatter plots are great for determining clusters, correlations, and outliers. Scatter plots require a continuous field in Rows (which produces the vertical or Y axis) and another continuous field in Columns (which produces the horizontal or X axis).

When you first plot the continuous measure fields in the Rows and Columns, you will see a single mark on your canvas.

You may be scratching your head and thinking this is not a scatter plot. You are correct, it is not, because it is missing the scatter. However, at this point, Tableau is simply following your instructions, which is to display the aggregate of one measure in Rows and the aggregate of another measure in Columns. In our recipe, it is the SUM(Salary/Winnings $) and the SUM(Endorsement $). This is really just an X and Y coordinate.

To create the scatter plot, this one point needs to be scattered. This can be done by adding dimensions in the Marks card that can force the scatter. For example, adding the Player field to Detail in the Marks card will make Tableau represent one mark per player, and each mark is the SUM(Salary/Winnings $) and SUM(Endorsements $) for that player.

If you want each record from the data source to be presented as a mark in your graph, you can drag the unique identifier for that row into Detail, and that will force the scatter. Alternatively, you can go to the Analysis menu and uncheck Aggregate Measures.

Notice that when you uncheck Aggregate Measures, the measure pills in your Rows and Columns will now be in disaggregated format, that is, Salary/Winnings $ instead of SUM(Salary/Winnings $).

In this recipe, we also added Sport to Color in the Marks card, so that we can visually identify which points belong to which Sport. Tableau 10 also introduces the data highlighter, which you can activate by right-clicking on a pill you have used in your view and selecting Show Highlighter.

In previous versions, the color legend allows you to highlight the points in the view, based on the color you select from the legend. The data highlighter is an improvement to this, and allows you to highlight the points interactively on hover, based on values you specify, or values that match a pattern, on the highlighter card.

For example, when you click on the Highlight Sport field, the list of values will appear. As you hover over the values, the corresponding Sport in the Color property is highlighted, as well as all the points in the scatter plot that belong to that sport.

The highlighter, however, can also work with the other dimensions you have used in your canvas, not just the pill that is in Color. For example, you can show the highlighter for player. This allows you to search for players, and highlight them in the scatter plot.

To follow this recipe, open B05527_01 – STARTER.twbx. Use the worksheet called Heat Map, and connect to the MasterTable (FlooddataMasterListrev) data source.

The following are the steps to create the heat map:

Heat maps allow us to easily see values that stand out based on the color density. In our recipe, we created a grid of colors based on the number of fatalities caused by floods. Where the color is darkest is when the number of fatalities was the highest.

To create a heat map in Tableau, we simply need to identify the dimensions we want and place them into our Rows and/or Columns. We also need to place a measure onto the Color in the Marks card, and this determines the density of the color for that intersection or column.

By default, measures, when dropped onto Color in the Marks card, will produce a gradient of colors. While pretty, sometimes the color differences are not so discernible. To make the colors more easily identifiable, we can elect to step the colors.

Stepping the color means there will be a finite number of colors used, instead of a gradient of colors that has all the color progression from the palette. The challenge with gradient is it is not easy to see how different one light blue is from another light blue, for example.

Another common variation for heat maps is adding another measure to Size in the Marks card. When you do this, the heat map will now have a series of different sized squares, instead of having a uniform size grid. For example, the following chart is what we will get if we place the measure Displaced onto Size.

What other use cases can benefit from a heat map? Heat maps can be used to spot trends in different verticals—to spot when students register, promotions are redeemed, profit increases, traffic is light/heavy and so on. The following screenshot can help identify shopper behavior—whether there is any particular month and day combination that seems to be most popular to shoppers.

To follow this recipe, open B05527_01 – STARTER.twbx. Use the worksheet called Text Table, and connect to the HollywoodMostProfitableStories data source.

The following are the steps to create the crosstab:

A text table is simply a series of text values and numbers in a grid. This is a pretty common request for organizations that are very numbers-oriented, and have just started transitioning to a more visual culture.

To create a text table in Tableau, usually you identify the dimensions you want to place in your Rows and/or Columns. There are a couple ways to create the text table:

Both of these actions introduce the Measure Names and Measure Values pills in your visualization.

Measure Names, which is discrete, gets placed in your Rows or Columns as a header, and Measure Values gets placed onto the Text property in your Marks card:

Tableau places limitations on creating text tables. By default, for example, only six columns will display properly. If you have more than six columns, the leftmost columns get concatenated.

To adjust this, you need to go to the Analysis menu, then Table Layout, then Advanced. There you can adjust the number of rows and column labels.

Tableau still, however, restricts this number. The maximum you can use is sixteen (16). Ultimately, if you need more columns and want to display all the numbers in the table, but not necessarily support visual analysis, then perhaps Tableau isn't the tool that is meant to do that job. Tableau is a great tool for visual analysis and exploration, so it is best to use it as such.

To follow this recipe, open B05527_01 – STARTER.twbx. Use the worksheet called Highlight Table, and connect to the HollywoodMostProfitableStories data source.

In addition, this recipe requires the text table in the Creating a text table (Crosstab) recipe to be created. This text table is the starting point of this recipe.

The following are the steps to creating a highlight table:

You might like the heat map because it emphasizes the items that have the darkest or lightest colors, but don't like the fact that it doesn't have the numbers.

You might like text tables, but it requires that you look at the numbers closely and focus before you can make sense of it.

You're in luck, you can create a hybrid table. You can have the text table, and color the grids according to your selected measure. This hybrid table is called the highlight table.

There is a trick to creating highlight tables in Tableau. The first is to create a text table. See the Creating a text table recipe in this chapter for the steps. Once you have the text table, you can drop a measure onto Colors in the Marks card. This colors your text table with a gradient of colors, which honestly will make your text table very hard to read.

Once your table already has the gradient-colored text, the final step is to change the mark to a Square. This is the magic sauce that creates your highlight table.

To follow this recipe, open B05527_01 – STARTER.twbx. Use the worksheet called Area, and connect to the DOHMH New York City Restaurant data source.

The following are the steps to create the area chart for re-inspections:

Area charts look like line charts (or time series graphs) that have shaded areas. Area charts are also similar to stacked bar charts, where the areas are cumulative. It is best used when you want to show the cumulative value of the measures you are using. Area charts, however, bear the same risks as the stacked bar chart. Because individual areas are stacked, it is hard to compare the different areas because they don't all start in the same level.

Creating an area chart is similar to creating your line chart and your stacked bar chart. You create a time series graph first. By default, Tableau will create a time series graph (or line graph) first when you put a Date or Date & Time field in either Rows or Columns, and a Measure in the other.

Once you have the time series graph, you can change the mark to Area, and drop a dimension in Color in the Marks card to create the different colored areas.

I've often asked students in my training classes when they would usually use area charts, and majority of the time students mention budgeting. With budgeting, they care about what proportion over time goes to which categories, but they also care about the cumulative value of the budget.

To follow this recipe, open B05527_01 – STARTER.twbx. Use the worksheet called Pie, and connect to the Top Athlete Salaries (Global Sport Finances) data source.

The following are the steps to create a pie chart:

Tableau comes with a built-in mark called Pie. When you switch the mark to a Pie, a new property appears under the Marks card called Angle.

To create a pie chart, simply change the mark to a Pie. Drag a measure to Angle, and a discrete field to Color.

Often a pie chart is accompanied by a label containing what the pie is, and what percentage of the pie that slice is. Percent of Total is a readily available Quick Table Calculation.

To add this as a label, you need to add a measure to the Label in Marks card first. This will initially display the actual values of that measure. Once added, you can right-click on this measure and select Quick Table Calculation. This should expand the readily available quick table calculations, which would include Percent of Total.

Pie charts often get a bad reputation in the visual analytics world. One reason is because humans can't naturally compare angles (or pie slices) easily, especially if the pie slices are pretty close in size. Bar charts are often more effective in showing the difference in magnitudes, because all we need to look at are the edges.

I think part of the issue is also abuse. Sometimes there is a tendency to overuse pie charts in presentations—especially the 3D tilted, exploded ones. Comparing angles is hard enough, tilting and exploding the pies do not help clarify the numbers.

However, pie charts can be effective in showing what proportion of the whole is allocated to something. I think pie charts have their place in the visual analytics and reporting world. Just like any other tools in your toolset, we just need to use them in the right place, right time.

To follow this recipe, open B05527_01 – STARTER.twbx. Use the worksheet called Bubble, and connect to the Data (Modified Gapminder Population) data source.

The following are the steps to create a bubble chart:

A bubble chart typically shows information in circles of varying colors and sizes. Creating a bubble chart in Tableau requires changing the mark type to Circle. A measure is typically placed in Size, as well as another dimension in either Color, Label, or Detail (or any combination thereof).

In this recipe, we placed the Population measure onto Size, created different bubbles for every country by placing Country in Label, and color the circles by Region.

If the pie chart gets a bad reputation, bubbles charts do too (maybe more). If we humans cannot compare angles with ease, we are also generally not very good at comparing diameter and circumference. Is that circle 15% bigger than the previous circle?

Most would argue that bar charts would be more effective than bubble charts, and I mostly agree. However, as with the pie chart, I think bubble charts can be effective—depending on the intent and audience.

In my courses, I ask students to present their work to the rest of the class. The goals of the presentation are to present their findings, as well as to engage the audience. One of the charts that frequently captures the attention of the audience is the bubble chart. Sometimes this chart also garners the most questions—therefore more engagement.

I would also argue that typically a single chart is not a be-all and end-all—typically we complement this with other charts, dashboards, and storytelling. There is nothing that stops us from starting our story with a bubble chart to get the audience's initial attention, and once we have their attention, proceed to more details using dependable bar and line charts.

The storytelling piece is also key. We have to be effective storytellers—our charts are simply our props. With effective storytelling, any chart can be effective and exciting.

To follow this recipe, open B05527_01 – STARTER.twbx. Use the worksheet called Word Cloud, and connect to the TopBabyNamesByState data source.

The following are the steps to create a word cloud:

A word cloud can be created in Tableau if you have a field that contains the keywords or values you want displayed in the word cloud, as well as a measure that determines the size of that keyword.

In this recipe, we change the mark to Text, and display the Top Name in Text.

In addition, we vary the Size of the names based on the number of occurrences it has in our data source. We also color the names differently based on Gender which is recorded from our data source.

In our tooltip, we also display in how many states this name was recorded as a Top Name. Since we don't use State anywhere in the visualization, but want to use it in our tooltip, we can make this data field available either by dragging it to Tooltip or Detail.

When you right-click and drag a data field onto Rows, Columns, or any of the other shelves, a window with different options for values comes up. When you right-click and drag State, you are going to be prompted with exactly what about State you want to display.

In our recipe, we chose CNTD(State), which is the numeric value for the unique number of states related to the Top Name.

You have probably seen some word clouds in some of the websites or blog sites you visit. Although it is hard to measure just by looking at the word cloud how many more times a keyword occurs compared to another keyword, this may be an effective chart depending on the context, intent, and audience. Arguably, bar charts would be more effective if you want to clearly state the difference in magnitude. However, in some cases, the intent is to simply illustrate and highlight which keywords are often used or mentioned. Sometimes, the relative comparison may be sufficient.

To follow this recipe, open B05527_01 – STARTER.twbx. Use the worksheet called Tree Map, and connect to the Data (Modified Gapminder Population) data source.

The following are the steps to create the population tree map:

A tree map allows us to present a part-to-whole relationship, just like a pie chart, but by subdividing a rectangle into smaller areas rather than subdividing a circle into different angled slices.

In Tableau, when you place a measure onto Size and a dimension onto either Color, Label, or Detail—or any combination thereof—the rectangle is broken down into smaller rectangles.

In this scenario, Tableau actually automatically assigns the Square mark. Although under Marks you will see Automatic, the icon refers to a Square. If you leave this as Automatic, or change the Mark explicitly to Square, you will get the same tree map.

Pie charts get a bad reputation because angles and slices are not easy to compare. Bubble charts are also hard to compare, because it is also hard to compare diameters and circumferences. Guess what, tree maps also fall into this same category. It is also not easy nor natural for us humans to discern and compare one area to another. Can we tell how much bigger or smaller one rectangle is compared to another? However, as with the other charts, with the right intent, audience, and story, this could be an effective chart—either by itself, or as a complementary chart to another.