5 Replies Latest reply on May 11, 2018 12:57 AM by Simon Runc

# Dynamic group defined by parameter

Simple example of a bigger problem…

I have a list of cities with their population.

For each city, there is a list of cities I want to compare to.

There is a parameter to select the city I’m interested in comparing against. Picking a different city produces its own custom list of comparison cities. A comparison city is not unique to one single selected city though (e.g. New York and Los Angeles both have Dallas as a comparison city). Every city on the original list would have its own list of comparisons, but I’ve only built out the example for comparisons to New York, Los Angeles, and Chicago.

In the attached Tableau file, you can see the expected result if New York were the selected city, but the grouping is hard coded to its specific list of comparisons. Changing the city selection does nothing right now, but I would like it to generate the dynamic groups of "Selected City," "Comparison Cities," and "Other Cities."

Any ideas? Thanks

• ###### 1. Re: Dynamic group defined by parameter

So how do you determine which Cities are in the comparison group? If I change to LA, say, how would you determine which are in it's comparison group?

• ###### 2. Re: Dynamic group defined by parameter

It's completely arbitrary and follows no set of rules. The comparison group is set on a separate table [see second tab of Excel file], and is subject to change at any time (i.e. today we might want to compare NYC to LA, Dallas, and Chicago; but tomorrow we might change the comparisons table so that NYC compares to LA, Dallas, and Seattle).

• ###### 3. Re: Dynamic group defined by parameter

So this is a tricky one, and the final solution will (likely) require some duplication of the data. I have (I think) a working version here. Let me try and explain what I've done!

First I UNIONed the 2 tabs together...

This also generates a "Sheet" column which tells me which Tab each row came from

I then joined the City Metric back onto the Compare to City, as I needed to get the Populations for the comparison.

Next we need to create some formulas to pick up the right stuff!

[City Selected/Comp]

IF [Sheet] = 'City Metrics' THEN

IF [City] = [Selected City] THEN 'Selected City' END

ELSEIF [Sheet] = 'Comparatives' AND [Base City] = [Selected City] THEN 'Comparison City'

ELSE 'Rest of Cities'

END

This classifies the Cities as Selected (using the original city column) and then uses the [base city] to pick up the right comparisons

We can then use this to pick up the right population column (we now have 2...one from the Metrics and the other where we joined the metrics to the Comparison Cities column)

[Population (to Use)]

IF [City Selected/Comp] = 'Selected City' THEN [Population]

ELSE [Population (City Metrics1)]

END

And finally, just so we get the sorting correct

[City (to Use) Order]

IF [City (to Use)] = [Selected City] THEN 100000000 ELSE [Population (to Use)] END

I'm assuming no city will have more than 100M people!

We can then filter out the NULLs and 'Rest of Cities' from the [City Selected/Comp] and we're good to go.

with a slightly different data structure, this could be made a bit easier (basically having a Base City row in the Comparison Data) like so

We'd still get duplicate rows, but we can just do a single join, and also the formulas are bit simpler. I've left the solution as per the provided data, in case you don't have that freedom.

Hope that helps, and makes some sense!

1 of 1 people found this helpful
• ###### 4. Re: Dynamic group defined by parameter

This is fantastic, thank you so much!

I have to admit I was unfamiliar with the Union feature (and I'm still a little confused about it since it doesn't have the constraints a SQL union would) but I see what is happening and will be able to use a form of this scaled out to my real data set. If I'm understanding correctly, then for every measure I might have, I'll need to create a calculated field to pick up the value from the proper sheet. But that's a small price to pay to make this work well.

Thanks again

• ###### 5. Re: Dynamic group defined by parameter

Yes it's a pretty ugly solution...but it gets the job done which is the main thing. Yes using the data in it's current form, we do need all these calculations in order to determine which City column we need and which population.

Just to show you how a bit of data-modelling at the start can make life easier in the long run....Attached is a version where I've added in the extra rows of data (as per my previous post)

As you can see we need a lot fewer rows, we need a lot few calculations and we only have a single population column. We are still duplicating some rows (where a city is in 2+ compare groups), so the SUM of population is no longer "additive" over all the data (eg. Phoenix is in 2 compare groups so it has 2 rows, and so the SUM of population for Phoenix is double it's actual population), but as we only ever look at Phoenix filtered down to single Base City it's not an issue.