Lorna Brown returned this week to set another table calculation based challenge involving a line chart which ‘on click’ of a point, exposed the ability to ‘drill down’ to view a tabular view of the data ‘behind’ the point. This is classic Tableau in my opinion – show the summarised data (in the line chart) and with ‘drill down on demand’. Lorna added some additional features on the dashboard; hiding/showing filter controls to change how the data is displayed in the chart, and a back navigational button on the ‘detail’ list.

The areas I’m going to focus on in this blog are

• Setting up the parameters
• Defining the date to plot on the chart
• Restricting the data to the relevant years
• Defining the reference bands
• Colouring the marks
• Working out the date range for the tooltip
• Building the table
• Drill down from chart to table
• Un-highlight selected marks
• Hide/Show filter controls
• Add navigation button

Setting up the parameters

This challenge requires 3 parameters.

Select a Date

a string parameter containing the 2 options available (Date Submitted & Date Selected) for selection, which when displayed on the dashboard will be set to a single value list control (ie radio buttons)

Latest X Years

an integer parameter, defaulted to 3, which allows a range of values from 1 to 5.

NOTE – Ensure the step size is set to 1, as this is what allows the Show buttons option to be enabled when customising the Slider parameter control type.

STD

another integer parameter, defaulted to 1, that allows a range of values from 1 to 3

Defining the date to plot on the chart

The Select a Date parameter is used to switch the view between different dates in the data set. This means you can’t plot your chart based on date field that already exists in the data set. We have to create a new field that determines which date field to select based on the parameter

Date to Plot

DATE(DATETRUNC(‘week’, IIF([Select a Date]=’Date Submitted’,[Date sent to company], [Date received])))

The nested IIF statement, is basically saying, if the parameter is ‘Date Submitted’ then use the Date sent to company field, else use the Date received field. This is all wrapped within a DATETRUNC statement to reset all the dates to the 1st day of the week (since the requirement is to report at a weekly level).

Note – there was some confusion which field the parameter option should map to. I have chosen the above, but you may see solutions with the opposite. Don’t get hung up on this, as the principal of how this all works is most important.

Restricting the data to the relevant years

The requirement is to show ‘x’ years worth of data, where 1 year’s worth of data is the data associated to the latest year in the data set (ie from 01 Jan to latest date, rather than 12 months worth of data). So to start with I calculated, rather than hardcoded, the maximum year in the data

Year of Latest Date

YEAR({MAX([Date to Plot])})

Then I could work out which dates I wanted via

Dates to Include

[Date to Plot]>= MAKEDATE([Year of Latest Date] – ([Latest X Years]-1),1,1)

In the MAKEDATE function, I’m building a date that is the 1st Jan of the relevant based on how many years we need to show.

So if Year of Latest Date is 2020 and Latest X Years =1 then Year of Latest Date – (Latest X Years -1) = 2020 – (1-1) = 2020 – 0 = 2020. So we’re looking for dates >= 01 Jan 2020.

So if Year of Latest Date is 2020 and Latest X Years =3 then Year of Latest Date – (Latest X Years -1) = 2020 – (3-1) = 2020 – 2 = 2018. So we’re looking for dates >= 01 Jan 2018.

This field is added to the Filter shelf and set to true.

So at this point, our basic chart can be built as

• Year on Columns (where Year = YEAR([Date to Plot])), and allows the Year header to display at the top
• Date to Plot on Columns, set to Week Number display via the pill dropdown option, and also set to be discrete (blue pill). This field is ultimately hidden on the display.
• Number of Complaints on Rows (where Number of Complaints = COUNT([XXXX.csv], the auto generated field relating to the name of the datasource).

To get the line and the circles displayed, this needs to become a dual axis chart by duplicating the Number of Complaints measure on the Rows, synchronising the axis and setting one instance to be a line mark type, and the other a circle.

Defining the reference bands

The reference bands are based on the number of standard deviations away from the mean/ average value per year.

Avg Complaints Per Year

WINDOW_AVG([Number of Complaints])

Once we have the average, we need to define and upper and lower limit based on the standard deviations

Upper Limit

[Avg Complaints Per Year] + (WINDOW_STDEV([Number of Complaints]) * [STD])

Lower Limit

[Avg Complaints Per Year] + (WINDOW_STDEV([Number of Complaints]) * -1 * [STD])

Add both these fields to the Detail shelf of the chart viz (at the All marks card level) and set the table calculation of each field to Compute By Date to Plot

This ‘squashes’ everything up a bit, but we’ll deal with that later.

Add a Reference Band (right click on axis – > Add Reference Line) that ranges from the Lower Limit to Upper Limit.

If an Average Line also appears on the display, then remove it, by right clicking on the axis -> Remove Reference Line – > Average

Colouring the marks

I created a boolean field based on whether the Number of Complaints is within the Upper Limit and Lower Limit

Within STD Limits?

[Number of Complaints]<[Upper Limit] AND [Number of Complaints]>[Lower Limit]

Add this to the Colour shelf of the circle mark type, and set to Compute Using Date to Plot. The values will be True, False or Null. Right click on the Null option in the Colour Legend, and select Exclude. This will add the Within STD Limits? to the Filter shelf, and the chart will revert back to how it was. Adjust the colours accordingly.

The Tooltip doesn’t show true or false though, so I had another field to use on that

In or Out of Upper or Lower Limits?

If [Within STD Limits?] THEN ‘In’ ELSE ‘Out’ END

Working out the date range for the tooltip

The Tooltip shows the start and end of the dates within the week. I simply built 2 calculated fields to show this.

Date From

[Date to Plot]

This 2nd instance is required as Date to Plot is formatted to dd mmm yyyy format and also used in the Tooltip. Whereas Date From is displayed in a dd/mm/yyyy format.

Date To

DATE(DATEADD(‘day’, 6, [Date to Plot]))

Just add 6 days to the 1st day of the week.

Building the table

Create a new sheet and add all the relevant columns required to Rows in the required order. For the last column, Company response to consumer, add that to the Text shelf instead (to replace the usual ‘Abc’ text). The in the Columns shelf, double click and type in ‘Company response to consumer’ which creates a ‘fake’ column heading. Format all the text etc to make it all look the same.

Add the Dates to include = true filter.

Also add the WEEK(Date to Plot) field to the Rows shelf, as a blue discrete field (exactly the same format as on the line chart). But hide this field (uncheck Show Header). This is the key linking field from the chart to the detail.

Drill down from chart to table

Create one dashboard (Chart DB) that displays the chart viz. And another dashboard that displays the table detail (Table DB). On the Chart dashboard, add a Filter Dashboard Action (Dashboard menu -> Actions -> Add Action -> Filter), that starts from the Chart sheet, runs as a Menu option, and targets the Detail sheet on the Detail dashboard. Set the action to exclude all values when no selection has been made. Name the action Click to Show Details

On the line chart, if you now click a point on the chart, the tooltip will display, along with a link, which when clicked on, will then take you to the Detail dashboard and present you with the list of complaints. The number of rows displayed should match the number you clicked on

Un-highlight selected marks

What you might also notice, is when you click on a point on the chart, the other marks will all ‘fade’ out, leaving just the one you selected highlighted. It’s not always desirable for this to happen. To prevent this, create a new field called Dummy which just contains the text ‘Dummy’. Add this onto the Detail shelf of the All marks card on the chart viz.

Then on the chart dashboard, add another dashboard action, but this time choose a highlight action. Set the action to run on select and set the source & target sheets to be the same sheet on the same dashboard. But target highlighting to selected fields, and select the Dummy field only

Hide/Show filter controls

Check out this post by The Data School that explains very simply how to work with floating containers to show/hide buttons. When creating in Desktop, the ‘onclick’ interactivity won’t work, you’ll have to manually select to show and hide, but once published to Tableau Public, it’ll behave as desired.

You have options to customise what the button looks like when the container contents are hidden, and what it looks like when they’re shown, via the Button Appearance

Add Navigation Button

On the Detail dashboard, simply add a Navigation object to the dashboard

and edit the settings to navigate back to the chart dashboard, as well as customise the appearance

Hopefully I’ve covered all the key features of this challenge. My published viz is here.

Happy vizzin’! Stay safe!

Donna