How to best visualize this data?
August 23, 2021 8:49 AM Subscribe
Looking for ideas on how to best visualize some data showing attendees at an annual community stakeholder event over time... Snowflakes inside.
The group of researchers I work with host a community stakeholder event every year, and we are trying to figure out how to best depict the flow of attendance at this event since its inception. Specifically, we are trying to show, for each year, the n of attendees who are new, those who have attended 1 prior year, 2 prior years, and so on.
Does anyone have any good ideas for how to visualize this? We could use stacked bars for each year, but it would be nice to have a more cohesive visual showing the flow. The overall picture across time is what we ideally want readers to understand; the exact n of attendees who have attended x number of times in any given year will be of less interest to everyone.
We thought maybe a Sankey diagram, but when my colleague looked into that it doesn't appear that it will work (it doesn't seem to handle "new" people, so is better for showing flow over time from a single startpoint).
Any ideas? =)
The group of researchers I work with host a community stakeholder event every year, and we are trying to figure out how to best depict the flow of attendance at this event since its inception. Specifically, we are trying to show, for each year, the n of attendees who are new, those who have attended 1 prior year, 2 prior years, and so on.
Does anyone have any good ideas for how to visualize this? We could use stacked bars for each year, but it would be nice to have a more cohesive visual showing the flow. The overall picture across time is what we ideally want readers to understand; the exact n of attendees who have attended x number of times in any given year will be of less interest to everyone.
We thought maybe a Sankey diagram, but when my colleague looked into that it doesn't appear that it will work (it doesn't seem to handle "new" people, so is better for showing flow over time from a single startpoint).
Any ideas? =)
While it would look better with continuous data, perhaps something like this? Put year on the bottom, years of attendance on the y-axis, have the total attendance be either some sort of cap on the top of the bars or a separate line.
posted by Jobst at 10:17 AM on August 23, 2021
posted by Jobst at 10:17 AM on August 23, 2021
You might be able to get your data to fit a cohort analysis. It's going to be hard to beat a stacked bar chart in terms of clarity though.
posted by TurnKey at 10:29 AM on August 23, 2021 [1 favorite]
posted by TurnKey at 10:29 AM on August 23, 2021 [1 favorite]
A Sankey diagram sounds right for this. You can make the new people clearer by coloring them differently, or having the line curve up from the bottom, rather than come in from the side.
You could also do a stacked bar chart, with the parts of the bar colored by the first year each attendee attended, which I think would be easier to read at first glance, but contain less overall information.
posted by wesleyac at 10:48 AM on August 23, 2021
You could also do a stacked bar chart, with the parts of the bar colored by the first year each attendee attended, which I think would be easier to read at first glance, but contain less overall information.
posted by wesleyac at 10:48 AM on August 23, 2021
Response by poster: Okay, I'm thinking we are just not understanding correctly how to fit our data into a Sankey. Can someone walk me through what this should look like in terms of nodes and flows?
posted by DTMFA at 11:13 AM on August 23, 2021
posted by DTMFA at 11:13 AM on August 23, 2021
This sounds like what you're doing: https://community.tableau.com/s/question/0D54T00000C5uSvSAJ/best-approach-to-create-a-sankey-where-items-enter-and-exit-the-flow-at-each-node
posted by bashing rocks together at 12:11 PM on August 23, 2021 [1 favorite]
posted by bashing rocks together at 12:11 PM on August 23, 2021 [1 favorite]
Response by poster: Yep that looks like what we want, bashing rocks together! Thanks, I'll read through that thread. =)
posted by DTMFA at 12:22 PM on August 23, 2021
posted by DTMFA at 12:22 PM on August 23, 2021
« Older Software and services for tracking news and... | The winter cometh and I need a new blanket Newer »
This thread is closed to new comments.
- A ridgeline plot could be used to show populations changing over time.
- A streamgraph might work, and seems a natural fit for this, but I feel like those give a very impressionistic view of the data.
posted by adamrice at 9:03 AM on August 23, 2021