Tag Archives: union

Creating a Dynamic Range “Parameter” in Tableau

Tableau’s native parameters have a couple of key limitations as of this writing in February 2018: The list of values is static and we can only select one value at a time, not a range of values. So when we want to do something like set up a viz with user-selectable top and bottom thresholds we have to set two parameters, not one, when what we really want is a nice range control like the view below:

What it looks like here is that I’m using a ranged “parameter” to change the mark color & shape whose set of possible values is based on other filters…which is what we’ve wanted in Tableau for awhile now. Now you might be asking “Why go to all this trouble, why not use Tableau’s JavaScript API or the (forthcoming) Extensions API?”. And my response to that is that not everyone has the expertise available to use Tableau’s APIs or D3 so we’ve got to make the best use of the tool that we have at hand. Curious about how can you build this for yourself? Read on to find out!

Continue reading

Using a Filter Action as a Parameter

I got a question recently about wanting to use an action in Tableau to set a parameter. For example in this view below the goal is to hover over a one of the bars below to send the action to the circles on top and use that value to color the marks, change the shape, etc. In this case what we want to do is some sort of evaluation like [Circles Continent] = [Selected bars Continent] to be able to flag the selected continent and treat it differently, just like we would if we had something like [Circles Continent] = [Continent Parameter].

But for actions that cross worksheets all we have are highlight & filter actions. Here’s what happens with a highlight action:

Tableau’s highlight is limited to greying out the non-highlighted marks and being able to optionally display text.

And if we try a filter action we are even more stuck:

The filter action removes all marks but the selected mark which then breaks the rank table calculation, positions the mark in the wrong place, and doesn’t really let us do things like change the selected mark’s color vs. the other marks.

So Tableau doesn’t have an action to set a parameter value so we’re kind of out of luck…or are we? Try out this viz, you can hover and see the color and size change while the rank value is still preserved:

Now you can get this kind of effect using Tableau’s JavaScript API, this was done without using any JS at all. Read on for an explanation of how you can do this for yourself! Also thanks to Rody Zakovich, he gave some feedback to this and came up with some great extensions that he’ll be posting about!

A caution for Tableau newbies: this uses some relatively advanced data preparation, Level of Detail expressions, data blending, filter actions, understanding of the difference between the grain of the data and the viz level of detail, and knowledge of Tableau’s order of operations. If those terms don’t mean anything to you then you might want to start out by learning about those first.

Tableau is a Data-Driven Drawing Engine

The key to all of this is that fundamentally Tableau is a data-driven drawing engine. By that I mean that what we see in the viz and available interactivity are dependent on the data. So if we feed the right data to Tableau we can get it to do (al)most anything we want. For example in a post from last year I set up waffle or unit charts inside a map.

In this case what we’re wanting to do is change Tableau’s interaction behavior across worksheets. Looking at our options for interacting across worksheets in a workbook:

  • Highlight actions can identify specific values have a very specific set of behaviors around appearance so we can’t change that.
  • Filter actions can identify specific values in the target viz but remove other values.

So there’s a loophole in filter actions…filter actions remove other values, but since the origin & targets of the filter action are coming from the data if we feed Tableau the right data we can have it keep what we want and no more. So in this case we just need to feed Tableau more data (as in copies of the data) so that after the filter action takes effect we have enough data to identify the selected & non-selected marks. Here’s a description of what I mean:

How I think of this is that we’re starting out with i continent values and what we’ll do is expand that out to some number j continent values (actually 2i or i*i), then the filter action will cut that number down to a manageable number k continent values that we can then use calculations to identify the selected and non-selected marks.

Introducing the Scaffold

A scaffold is used in building construction, and in Tableau a scaffold data source is one that helps us get the data “just so”. In Multiple Ways to Multi-Select and Highlight I did a version of this where a union was used to give enough data so that way a mark could be highlighted. That required a full union of the data which can get prohibitively large, so for this method we’ll use a scaffold source that has just the values we need, and then when we want measures from the underlying data we can use a Tableau data blend.

The scaffold uses multiple copies of the list of the values that we want to filter for (Continent in this case). Now if you just have a “flat” table of data and don’t have a separate unique list of values then there are multiple ways to get one, please see Creating a List of Values in Tableau from Text and Excel Sources. I’ll be using

Once you have the list there are two different scaffolds you can use: One uses a cross product (i.e. for every continent there is every other continent), the second uses a union (thanks to Rody for that suggestion and demo). I’ll go through the cross product scaffold first because that’s a bit easier to set up than the union.

Using a Cross Product Scaffold to use a Filter Action as a Parameter

This section goes through the cross product scaffold. A cross product is also called a cross join or cartesian join or Cartesian product and a simple description is “for each value of A return each and every value of B”. So if we start out with the two values [A1, A2] and three values [B1, B2, B3] then we get the six combinations [A1B1, A1B2, A1B3, A2B1, A2B2, A2B3].

In this case we’re building a cross product of the dimension we want to build an action on and for this example we’re using Continent so the cross product will be 6 continents * 6 continents and end up with 36 rows in the scaffold. It’s important that the scaffold only has one record for each combination, if it has more than one record then the calculations below will break and alternative formulae would be required.

I’ll explain a little further about how this ends up working down below.

Creating the scaffold and setting up initial interactivity

  1. In Tableau connect to your original data source, in this case Superstore.
  2. Followed the instructions for creating an aggregated extract source from  Creating Lists of Values for Tableau from Text & Excel Sources. For this next step I used the Continent dimension. Note where you saved the extract.
  3. Connect to the extract (.tde or .hyper file) in Tableau.
  4. Drag a second copy of the extract onto the canvas.
  5. In the join window set up an inner join with two join calculations so the join is 1 = 1.
  6. Rename the copy of the dimension to something useful, I used xprod continent. (xprod is short for cross product).
  7. Rename the data source to something useful, I named it xprod Continents.
  8. Create a worksheet for the target using the scaffold (xprod) source as the primary with any necessary fields from the secondary source. This view requires the dimension & xprod dimension to be somewhere on the viz. To help see what’s going on I used a crosstab to start. Note that the xprod dimension is not in the compute using of the rank table calculation since there are multiple copies of the data.
  9. Connect to your original data source.
  10. Create the origin worksheet, in this case it’s a simple set of bars:
  11. Build a dashboard with the origin and target sheets.
  12. Add a filter action as a Select filter and and add a filter that for the source field uses the original dimension from the raw source and for the target field uses the xprod dimension from the scaffold source from step 6.

Here’s a demo:

How does this work?

The scaffold source has 6 copies of the data, one for each xprod Continent. The filter action targets the xprod continent so when the filter action is triggered only one xprod Continent remains, and because we’ve multiplied the data there are the 6 Continent values remaining.

This leaves us with two useful attributes for each mark – the Continent, and the xprod Continent that identifies the selected value. The xprod continent is effectively the [Selected bars Continent] or [Continent Parameter] that we originally wanted to be able to do evaluations like  [Circles Continent] = [Selected bars Continent] or [Circles Continent] = [Continent Parameter], only we had to do some extra data preparation to get there!

Identifying Selected Marks for Setting Color, Size, etc.

Now to we can do the evaluations to identify the selection status. There are three states to track:

  • No selection made at all (which is something we can’t do with a regular parameter)
  • The selected mark
  • The non-selected marks (when there is a selected mark)

When there is no selection at all then there are 6 xprod continents for each Continent so we can count those and be able to flag the selected/non-selected state. Then if a selection is made the Filter Action reduces the data to only one value of xprod continent so we can test for that to identify the selected mark vs. non selected marks.

Here’s the Selection Status (xprod) formula used in the scaffold source:

//given the scaffold source COUNT(continent) across the data will return more than
//1 when the scaffold isn't filtered
//this uses the ability of EXCLUDE LOD expressions to be evaluated as
//record level calcs before they are aggregated in the view
IF {EXCLUDE [xprod continent]: COUNT([Continent])} > 1 THEN
    -1 //no continent selected
ELSE
    // identify selected continent
    IF [Continent] = [xprod continent] THEN
        1 //selected
    ELSE
        0 //not selected
   END
END

With this flag now in place we can create additional calculations that can be placed on Color, Size, Shape, Label, Tooltip, etc. or even elsewhere in the viz.

For example here’s the Highlight Text calculation:

IF [Selection Status Flag (xprod)] = 1 THEN "I'm selected!" END

This only returns “I’m Selected” for the selected mark and Null for everything else. By putting this on Label it only appears when the mark is selected and can be used on Color as well. Note that it uses the ATTR() aggregation because the Selection Status Flag (xprod) is using an EXCLUDE LOD expression.

I created another calculated field for Size and some customization of the Size so that the nothing selected state has a mid-size neutral state, the selected mark is large, the non-selected marks are small. Here’s the completed viz:

Setting up with Select and Iterating

A couple of notes on setting these calculations up in the view – since we are using fields that have different results depending on the filter action status we will need to do an iterative process. For example when using Highlight Text on Color I needed to put the field in, trigger the filter action as a select action (so it would stay in place when I moved off the mark), then set the color for the selected mark, then verify everything was working by turning the action on and off, and then finally making the action a hover action.

Removing the Extraneous Scaffolded Marks

If we select all the marks in an unfiltered/non-selected scaffold view there are 36 marks – behind each Continent mark we can see there are the scaffolded continent marks from the xprod continent dimension. Personally I don’t like views that have extra marks kicking around for the following reasons:

  • The more marks Tableau has to draw the slower the viz.
  • Even though the marks are hidden they can cause confusion on the parts of users as they interact with the viz.
  • The extra marks will be part of any viewing of summary data or data downloads and that can be especially confusing.

So how can we filter out the extraneous marks? This is where knowledge of Tableau’s order of operations, the viz level of detail (vizLOD), and filter actions comes into play. ~~link to documentation. The vizLOD is Continent and xprod continent and when there’s no filter action there are 6 xprod continents for each Continent, whereas when the filter action is activated there is 1 xprod continent for each Continent. Now the filter action is applied as a record-level aka dimension filter in Tableau’s order of operations ~~link so we need to use a filter that comes after that which could be an aggregate filter, a table calculation filter, or (as in this case) and INCLUDE or EXCLUDE LOD expression-based filter. Here’s the formula for the Remove Extra Marks (xprod) calculation:

[xprod continent] = {EXCLUDE [Continent], [xprod continent] : MIN([xprod continent])}

This uses a variation of the technique from my earlier post on identifying a dimension at a lower level  where we’re using a Level of Detail expression to compute a result as an aggregate and then comparing it a record level. In the filtered view we know there’s only one xprod continent for each continent so that works out just fine.

~pic of selected

In the unfiltered view the EXCLUDE LOD will return the first xprod continent (probably Antarctica) and then only that one is kept while showing the 6 continents.

~pic of not selected

With this filter in place we end up with only 6 marks either way and have removed the extra marks added by the scaffolding to get a nice clean viz.

Final Notes on using a Cross Product Scaffold to use a Filter Action  as a Parameter

This is not a technique for the faint of heart, it’s using a wide range of Tableau’s functionality to get a specific set of user interactivity. So it might not be for you. In building views like this for me where I’ve worked out the details of how the calcs need to work the most challenging part is often building the scaffold source. For example if you have hundreds or thousands of values of the dimension(s) you need to scaffold then the cross product can get prohibitively large, and for that we’ve got the alternative of using a union, we’ll cover that in the next section.

Using a Union Scaffold to use a Filter Action as a Parameter

Rody pointed this out to me as an option, this method uses a union’ed scaffold source instead of a cross product and a filter action whose filter pill is set to Exclude. So the scaffold source can be a lot smaller, but the set up is a little more complicated.

Overview 

For this method instead of having N sets of values in the scaffold there are only 2 sets of values. We set up special calculated fields in the scaffold and the original data that will enable the filter action to exclude (remove) from a selected value from the scaffold so we can use that difference to detect what has been selected.

How to Build the Union Scaffold

Here’s how to build this, this is a slight variation on the instructions for an aggregated extract from Creating Lists of Values for Tableau from Text & Excel Sources:

  1. In Tableau connect to the raw data source.
  2. Union the raw data to itself.
  3. Create a worksheet that only has the necessary dimensions plus the Table Name and Sheet dimensions.
  4. Create an aggregated extract per the instructions in the link. ~pic

This ends up with a scaffold source where there are two copies of the list of values, like this: ~pic

Setting Up Interactivity

  1. In the original data source create an ExcludeOrigin field in the original data with the formula '~~' + [Continent].
  2. Create an origin worksheet with the Continent & ExcludeOrigin fields.
  3. In the scaffold source create an ExcludeTarget field with the formula:
//there's an implied ELSE Null, the Null values are the ones we will ultimately keep
IF [Table Name] = 'Data1' THEN '~~' + [Continent] END
  1. Build the target Scaffold sheet with Continent and ExcludeTarget as dimensions. Note that there are 2N marks where N is the number of Continents with 2 values of ExcludeTarget for each. ~pic
  2. Add any measure(s) you want from the original data via a data blend.
  3. Create a dashboard with the two worksheets.
  4. Add a filter action on Select from the origin worksheet to the target worksheet that goes from the ExcludeOrigin field to the ExcludeTarget field. ~pic
  5. Trigger the filter action by selecting a mark on the origin worksheet.
  6. Go to the target worksheet.
  7. Right-click on the Action (ExcludeTarget) pill on Filters and choose Edit Filter…  The Edit Filter window appears. (If you don’t see the pill on Filters  then you haven’t triggered the Filter Action).
  8. Click on Exclude, then click OK. ~pic
  9. Go back to the dashboard and click on different marks on the origin worksheet, you’ll see the target update.

To explain how this works we have to keep in mind that there are effectively two states:

  • When there are no marks selected in the origin worksheet then nothing is excluded from the target sheet and we see all N (12) marks from the scaffold.
  • When a mark is selected in the origin worksheet then the corresponding mark with the non-Null value of ExlcudeTarget is removed from the viz, leaving us with N-1 (11) marks remaining.

Identfying Selected Marks for Color, Size, etc.

Because this scaffold is built using a union the detection of mark selection status works a little differently, here’s the formula for the Mark Selection Status (union) field:

//In the union scaffold there are two states: all rows exist or one has been filtered out by the selection
//if all rows exist then there are 2x the number of continents and we can test for that
IF {EXCLUDE [Continent], [ExcludeTarget] : COUNT([Continent])} % 2 = 0 THEN
    -1 //no selection made
ELSE
    IF {EXCLUDE [ExcludeTarget] : COUNT([Continent])}  = 1 THEN
        1 //the selected value
    ELSE
        0 //the non-selected values
    END
END

Essentially since we’ve doubled the data then we can use the modulo (%)  operator to detect that doubling and identify the no selection status, then by counting continents we can find out whether are 1 or 2 records and identify the selected/non-selected marks.

From here the other calculations are all the same as for the cross product scaffold except for the Remove Extra Marks calculation. In that case the Remove Extra Marks (union) formula is:

ISNULL(ATTR([ExcludeTarget]))

Note that we could just use ATTR([Exclude Target]) and filter for Null as an alternative…this is one of those cases where I like having a separate calculated field because then by the name of the field I can give the viz maintainers a chance to understand what is going on.

Here’s a completed dashboard using the union, you’ll find the interactivity to be the same as the cross product version:

Conclusion…or…When Should I Use This?

When I’m building a dashboard and my users are wanting interactivity that is more than what Tableau immediately offers I go through a mental checklist:

  1. Is the goal something that we can pull off using highlighting, sheet swapping, filter actions, parameters, sets, etc.?
  2. Is this the only “ask” for additional interactivity or are there other cases for this dashboard where the desired user experience is pushing the boundaries of what is provided in Tableau? If so are there resources to use some JavaScript and Tableaus JS API?
  3. Only then do I start considering more complicated methods that require more data prep and configuration like the one presented here.

Here’s a link to the Filter Action as Parameter dashboard on Tableau Public. Hopefully you learned a bit about how to take advantage of Tableau’s capabilities, if you have any alternatives or questions please ask in the comments below!

Creating Lists of Values for Tableau from Text & Excel Sources

There are various use cases where we start out with a “flat” table like the Superstore sample data that has a number of columns with various dimensions and we want to make a simple list of unique values of one or more dimensions. such as a list that has just the six continents in Superstore:

The use cases for this include:

  • Using a filter action value as a parameter in the target source (look for posts from myself and Rody Zakovich on this in the next week).
  • Cross data source filters with higher performance when the list of filter values can be small compared to the volume of data.
  • Creating scaffold data sources to pad out data and ensure there are no sparse combinations of values.
  • Situations where we’d want to do a union or cross product of the data to do something like a market basket analysis but the union or cross product would be prohibitively large, so instead we only union or cross product desired dimension(s) and then join in the original data as necessary.
  • The last multi-select highlighter method from Multiple Ways to Multi-Select and Highlight in Tableau can use a self-union.

If you are starting out with a well-structured data warehouse with dimension tables, can write SQL, Python, or R, build custom views on the data source, use data preparation tools like Alteryx or Easymorph or Trifacta, etc. then obtaining or generating these kinds of lists is pretty straightforward. But not everyone has those skills or resources, and in the case of users who just have Excel and/or text files we need to get creative. This post goes through a three different methods to get these lists in Tableau:

    1. Ask!
    2. Aggregated Extract
    3. Excel Pivot Table as a Data Source
    4. Custom SQL

In this post I’ll go through each of these options. [Note: this post was updated on 10 Jan 2018 to make the aggregated extract method a little simpler.]

1. Ask!

This might seem obvious, but sometimes we’re stressed out and under deadlines and don’t realize we might be able to get help. If the data you are working with is coming from someone else then go ahead and ask them if they have a list of unique values. I’ve found that most people want the data they produce to be used and used well and if I’m coming back to them asking for something so I can do more with “their” data they are happy to accommodate me. I might phrase the request like “I want to make sure I’m using the latest list of departments, can you give me that list?”

The one caveat to getting data back from your ask is that you’ll need to go through some validation to make sure the list matches up with the “real” data, sometimes the amount of validation and cleansing isn’t worth the effort and one of these other approaches is better. However if you’re in a data-starved environment the kind of relationships you can make by asking for data can lead to more trust and ultimately more access to the data you want (and need).

2. Aggregated Extracts

For this method we’re going to connect to the data source and build an extract only we’ll be telling Tableau to aggregate the data to the desired level of detail (the field(s) we want to use) before Tableau builds the extract. The resulting extract then just has a record for each combination of field(s) that we want to use.

  1. Connect to the data source.
  2. Create a single worksheet with the field(s) you want to use as dimension pills, I usually just put them on Rows as discrete (blue) pills:
  3. Right-click on the source and choose Extract Data… The Extract Data window opens.
  4. Click on the Aggregate data for visible dimensions checkbox.
  5. Click the Hide All Unused Fields button.
  6. Click Extract. Tableau will ask where to save the extract. Choose a location and click OK.

Voila, you now have an aggregated extract source that you can use in Tableau data blends and/or join to!

Notes on Aggregated Extracts

There are a few things to keep in mind when using aggregated extracts: First of all there’s the need to refresh them to keep up with the data so if you have Tableau Server you’ll need to set up an appropriate schedule, if not then you’ll need to set up your own manual or automated workflow that gets the results you need. One possibility is using Tableau’s extract API.

Secondly if new columns are later added to the data they are automatically added to the extract. This may be ok for some use cases, there are others where this will break views that depend on that extracted data.

Finally, if you want to join on this aggregated extract you’ll need to join directly to the .tde or .hyper file.  Where this gets complicated is handling data updates. You’ll need one workbook or workflow to update the extract and then use the extract in a second workbook. Unfortunately we can’t publish the extract to Tableau Server or Online and join to that published data source (yet), otherwise that would be an easy workaround. There are a number of cases where a Tableau data blend is sufficient, we’ll be demonstrating one in the next week.

3. Excel Pivot Table as a Data Source

For Excel sources besides connecting to worksheets with raw data we can connect to worksheets that are built as a pivot table.

Here’s how using Excel 2016 for Mac:
  1. Open the source in Excel.
  2. Create a pivot table in a new worksheet.
  3. Drag the field(s) you are interested in to Rows.
  4. Rename the Row Labels header to have appropriate values if necessary.
  5. Remove the grand total.
  6. Rename the worksheet to something more meaningful than Sheet2.
  7. Save the workbook in Excel.
  8. Open up Tableau and connect to the Excel workbook.
  9. Drag the pivot table you just added onto the canvas:
Now you can use this to join to other tables and/or use in data blends.

Notes on using Excel Pivot Tables as a Data Source

Before Tableau introduced Level of Detail expressions in version 9 I used pivot tables in production views to pre-aggregate the data for some values and also to create tables I could join on to pad out the data so I could be sure to see records for every (person, office, metric) for every month. This method has one potentially major challenge around data updates, though, and that is that if we have data in worksheet A and a pivot table in worksheet B and we update the data in A (such as adding a new value that should appear in the pivot table B) that change won’t be reflected in the pivot table B until there is an explicit command in Excel to update the pivot table B and then save the workbook.

Even though we can tell Excel to do things like “Refresh data when opening file” this flag is only detected by Excel, not Tableau. Therefore to get updates to the data to be reflected in the pivot table the workflow has to include the steps to do a Data->Refresh All or open the pivot table worksheet before saving the workbook.

4. Custom SQL for Excel & Text Files

When I’m delivering Tableau training classes and we get to the point of talking about SQL & Tableau there are two common reactions: 1) yeay! and 2) [eyes glaze over]. This part is for the people in the latter category. Tableau hasn’t turned everything we might want to do into point & click, so sometimes we need to work with raw data. We do this in our everyday lives…there’s no good vegetarian restaurant in my town so when my wife & I want African ground nut stew we’ve got to make it ourselves. So I think of using Custom SQL as using the raw ingredients of the data to get a result I don’t have another way to get. However, in this case we’re going to be lazy (in a good way) and make Tableau write the SQL for us! Here’s how (these instructions don’t work for Tableau for Mac, see the Notes section below for more info):

    1. Start adding a new data source that is the Excel or text file you want to connect to.
    2. In the Open dialog select the file, then on the Open button click the drop down carat and choose “Open with Legacy Connection”.  You’ll return to the data source window.
    3. Drag the worksheet or file if necessary onto the canvas.
    4. Use the Data->Convert to Custom SQL menu option. The Convert to Custom SQL window will appear.
    5. Edit the Custom SQL to remove all the fields that you don’t need.
    6. Make sure to delete the trailing comma from the last field in the SELECT before the FROM.
    7. Add the DISTINCT keyword after the SELECT before the first field. The SQL query will now look something like this:
    8. Click Preview Results… to test. If it comes back with an error then check your syntax (see notes below for some tips) and try again. If it works by showing a View Data window with your results close the View Data window and then click OK to close the Custom SQL window You’ve now created a unique list of values using custom SQL!

The advantage of using Custom SQL compared to using an aggregated extract or pivot table is that it updates with the data and doesn’t require the more complicated workflows of the other methods.

Simple SQL SELECT Query Syntax

Here’s a really simple example for getting one field from one table:
SELECT DISTINCT [table].[field1] AS [field1]
FROM [table]
If you want multiple fields from one table the SQL query looks like this:
SELECT DISTINCT  [table].[field1] AS [field1],
   [table].[field2] AS [field2],
   [table].[field3] AS [field3]
FROM [table]

In some ways SQL is written a little backwards, and in more complicated queries backwards and forwards. To me the real “starting place” of a SQL query is the FROM part because that is telling the SQL engine where (what table, worksheet, or text file, generically called “table”) to get the data from. Then the SELECT is going to grab the set of fields that we specify. The DISTINCT keyword tells the SQL engine to only get the unique (distinct) combinations of values of those fields instead of grabbing every single record.

The field names themselves use the [table name].[field name] convention so that if there are multiple tables in a query each field referenced can be uniquely identified. The table and field names are surrounded by square brackets by default to handle situations where the table or field name might have spaces. Finally Tableau uses the AS [field name] aliasing option to ensure that the name used by Tableau is a usable name in Tableau.

SQL doesn’t care about spaces & line feeds, we could write SELECT DISTINCT [table].[field1] AS [field1] FROM [table] all one one line and it would work just fine.

SQL cares very much about the placement of square brackets & commas, if one is out of place or missing then the whole query will fail. Make sure that you have all brackets in place and make sure that the last field in the SELECT doesn’t have a comma after it.

Notes on Custom SQL for Excel & Text Files

The Legacy Connector is not available on Tableau for Mac, so we can’t use this particular method for connecting to Excel or text files on the Mac.

The Legacy Connector is actually the Microsoft JET driver that was phased out in Tableau version 8.3 for a variety of reasons, here’s a link of differences to be aware of from the Tableau legacy connector documentation. Also here’s the Tableau documentation on Connect to a Custom SQL Query. Finally I did a post awhile back on details of using the Custom SQL in the context of Microsoft Access connections which also use the MS JET driver, some of the points there are useful to keep in mind.

Hacky…or not?

If it all seems a bit hacky and contrived then I agree with you. At this time if all we have are Excel or text files and what features Tableau provides we’re in a low-resource environment and workarounds are necessary.

I regularly see projects I’m working with needing to invest more in data preparation in order to keep Tableau humming along. That investment could be in scripting languages like Python or PowerShell or R, using PowerQuery, starting the process of moving data into a database (there are free versions of many databases), and/or use more dedicated data preparation tools like Alteryx, Easymorph, or Trifacta. I like to set expectations around this early on in new projects because once they start using Tableau invariably projects run into imitations of their existing data pipeline to provide the volume and variety of data that they can now analyze in Tableau.

Conclusion

The goal for this post was to set you up with the skills you need to get a custom list of distinct values to support several different use cases and I hope this did that for you. As mentioned early on, Rody Zakovich and I have some posts in the works that use this to do some new things in Tableau!

Multiple Ways to Multi-Select and Highlight in Tableau

Last year in version 10.0 Tableau introduced the highlighter that lets us quickly highlight marks. It’s got two potential limitations, though:

  1. We can only highlight a single value or all marks meeting a search criteria, not multi-select values.
  2. Tableau’s Highlighter (and Highlight Actions) are hard-coded so we don’t have control over the formatting of the highlighted marks.

We can work around those limits in a few different ways:

  • Using a separate worksheet with Highlight Action(s) enabled with Hover or Select.
  • Using a separate worksheet with Highlight Action(s) enabled, a dual axis, and a duplicated dimension to display different marks. This technique was developed by Rody Zakovich in his post Only Color Marks on Dashboard Highlight .
  • Using a self-union’ed data source and a dual axis to get total control over how the highlighted marks are displayed and have more control over the user interface for choosing the highlighted marks. I think I might have invented this technique, I haven’t seen anything quite like it before. This method lets us build views like this:

Read on for how to build these out and choose the right method for you!

Continue reading

Cross Data Source Joins + Join on Calcs = UNION Almost Anything in Tableau v10.2

Since Tableau v9.0 or so every new release has come with new features that simplify and reduce the amount of data prep I have to do outside of Tableau. Pivot in version 9.0, the first batch of union support in v9.3, support for ad hoc groups in calculations, cross data source joins and filters in v10.0, and more in-database unions and join calculations in v10.2. With the join calculations we can now do unions and cross/cartesian joins within or across almost any data source without needing Custom SQL or linked databases and without waiting for Tableau to implement more union support, read on to learn how!

Here are some use cases for unions across data sources:

  • Union data that is coming from different systems, for example when different subsidiaries of an organization are using different databases but you want a single view of the company.
  • Union actual sales data from transactional systems and budget data that might come from an Excel spreadsheet.
  • Union customer & store/facility data sets so you can draw both on the same map.

This post goes through examples of all three using a combination of text files and superstore, and Rody Zakovich will be doing a post sometime soon on unions and joins with Tableau data extracts. (Did you know you could do cross data source joins to extracts? That capability came with v10.0, and we can have all sorts of fun with that using join calculations!)

Continue reading

Customizing Grand Totals – Part 3

In Part 1 and Part 2 of this series, I described how Tableau computes Grand Totals and several options for generating your own Grand Totals. In this post, you’ll learn the most flexible method for customizing Grand Totals, via custom SQL to duplicate the data.

In November 2013 Tableau 8.1 also added a new two pass totals feature that may remove the need for customizing grand totals for your use cases, more details are in Tableau 8.1 Two Pass Totals. In December 2015 Tableau 9.3 added total control for placing totals to the left and/or on top of the view.

Continue reading