Rank themes by frequency

Hello,

I would like to count and rank themes of concerns on my dashboard so we know which issues come up the most frequently. These are not pre-populated and will change over time as new themes come up.

Example:

In these 5 rows, I would want to display: Care plan (4), Communication (1), EVS Variation (1). If there were a tag of "cleanliness" next week, I would like the new list to autopopulate with: Care plan (4), Communication (1), EVS Variation (1), Cleanliness (1). Ideally, I would like my dashboard to only show the top 10 themes.

Any recommendations? I have thought about COUNTIF and HAS formulas, but cannot figure out how to include new themes as they emerge.

Thank you

Find more posts tagged with

Accepted answers

All comments

Sam_Harwart

How frequently will the themes change? If the list is known or wouldn't be difficult to update, there are some possible solutions.

If it needs to be 100% dynamic, you could use a report that summarizes & groups by count but that doesn't split multi-select columns or sort only the top 10.

It's an interesting problem. Looking forward to seeing what people can come up with to solve.

Sam_Harwart

Hey @Kate HC - still looking for a solution? I've got something that, albeit kind of messy, works. It uses a helper sheet and would need some manual intervention every so often to add rows, but is generally pretty automated. Let me know if it would be helpful and I can post details!

Kate HC

Thanks for the response. It would need to be as dynamic as possible. We have lots of responses to sort through, so manually updating would be very difficult.

I would appreciate your thoughts.

Sam_Harwart

@Kate HC

First I made a sheet that holds the data - I've used colors where you'll have your themes.

Everything else happens on the helper sheet:

The [Primary Column] is 0 - N, where N is the max number of combos you think you might realistically see. I used 50 for my testing, but there's no reason you couldn't set up way more.

The [Raw Uniques] parses the list of colors from the multi-select column on the original sheet and extracts all of the unique combinations. It creates a list of the combinations and then uses the [Primary Column] value to get the Nth item in the list. You can make this a column formula.

=IFERROR(INDEX(DISTINCT({Colors}), [Primary Column]@row), "")

[Combine Unique Multi] takes all the unique colors from the list of unique combinations and puts them into a single cell. You'd need to change the "50" value here if you want more rows. CHAR(10) is a return, which is what separates multi-select values.

=JOIN([Raw Uniques]2:[Raw Uniques]50, CHAR(10))

Next, I turned it into a normal text string, again using CHAR(10) in the [Convert to Text, What's Left] column. This is where some of the formulas are different for the first row or two.

For Row 1, I used this formula:

=JOIN([Combine Unique Multi]@row, ";")

For Row 2 and beyond, we're going to use "What's Left" after we remove each unique color as we go through the list.

=RIGHT([Convert to Text, What's Left]2, LEN([Convert to Text, What's Left]2) - [Find CHAR 10]@row)

[Find Char 10] is going to locate the next return separating each value in our list. Row 1 is blank, Row 2 has this:

=IF([Find CHAR 10]@row = 0, [Convert to Text, What's Left]@row, LEFT([Convert to Text, What's Left]1, [Find CHAR 10]@row))

Eventually, we get to the bottom of the list and don't find any other CHAR(10) values, so we'll have to deal with that as well.

[Separate Multi] extracts a unique value for each row, using the CHAR(10) as its delimiter.

=IF([Find CHAR 10]@row = 0, [Convert to Text, What's Left]@row, LEFT([Convert to Text, What's Left]1, [Find CHAR 10]@row))

Now we're getting to the easier part. [Color Frequency] looks at the original sheet to count the occurrences of each unique color.

=COUNTIF({Colors}, CONTAINS([Separate Multi]@row, @cell))

I couldn't find a better way to do this earlier in the process, so [Color Name] now eliminates the repeated final value. This is the one we'll use for our reports.

=IFERROR(INDEX(DISTINCT([Separate Multi]:[Separate Multi]), [Primary Column]@row), "")

From there, we can build a simple report using the [Color Name] and [Color Frequency]. Just filter out any error values from the formulas.

Now you could also add in a column that used =LARGE to rank the frequency of the colors and then only pull in the top 10 on your report, or pull in them all to a dashboard and then make it small enough that only the top XX show.

For example, I sized it to only show the top 5. You can scroll to see the rest if needed.

Whew. This was a mess and I am 1000% sure this isn't the best way to do this, but I've at least got it working. The computer science classes I took back in undergrad lead me to believe this is a super inefficient way to solve the problem and a python script would have made quick work of it.

Maybe someone like @Andrée Starå will come drop in and find a single formula to solve the whole thing 🤣

Let me know what you think and what questions you have!

Kate HC

Wow thank you for all your effort on this - It may take me a few days to sort through it! I will let you know what questions I have.

Sam_Harwart

Sure thing. It was a fun problem to try and solve. Hopefully it helps! I'm sure you'll come up with a more efficient solution as you try and implement it.

Sign in to join the conversation!

Quick Links

Component react.discussion.discussions had an error.

ref must be one of: categoryID, siteSectionID, category, category/categoryID, category/name, category/description, category/url, category/allowedDiscussionTypes, locale, siteSection, siteSection/basePath, siteSection/contentLocale, siteSection/sectionGroup, siteSection/sectionID, siteSection/name, siteSection/description, siteSection/apps, siteSection/attributes, layoutViewType, discussionID, commentID, page, sort, discussion, discussion/name, tags, breadcrumbs, discussionApiParams, serverDraftID, serverDraft.