Hi,
I have a table looking at how users return to a product over time. I can do this in columns for each week, error bars for each column - perfect.
I now want to look at subgroups of users that did or didn't do a specifc activity. I use the activity as an overlay, again no problem.
But now I'd like to compare different activities - which one(s) help most to boost retention.
I've created a single column that reports all of the activities as multiple response - most people only have one, but a few have several activities. When I use this variable in a graph, I get a seperate group for each possible combination - useless. I need a group for each activity, and yes, the groups will overlap, i.e. the same person show up multiple times.
the best i can do is create a dashboard that shows the individual graphs side by side, but that's still awkward to compare. I want just the "yes" bars together in one graph...
Any idea how I could do this?
Thanks!
Susanne
I'm not sure if this is what you are looking for, but I stacked the activity columns and it seems to be able to show the number of people engaged in each activity (even with overlaps). I believe stacking those columns is what you want to do. See attached (and script for an initial graph).
Hi, Susanne!
There are some automatic charts in the Categorical platform that might help you out. Here's a script with some basic ideas:
Categorical(
Grouping Option( Both ),
X( :Week, :Name( "Returned?" ) ),
Multiple Delimited( :Name( "Used any?" ) ),
Legend( 0 ),
SendToReport(
Dispatch(
{},
"Multiple Delimited( Used any? ) By Week*Returned?",
OutlineBox,
{Close( 1 )}
)
)
)
Here's a description of the charts:
http://www.jmp.com/support/help/The_Categorical_Report.shtml#134897
And a description of the "Multiple Delimited" responses:
http://www.jmp.com/support/help/Example_of_the_Categorical_Platform.shtml#129405
I'm not sure if this is what you are looking for, but I stacked the activity columns and it seems to be able to show the number of people engaged in each activity (even with overlaps). I believe stacking those columns is what you want to do. See attached (and script for an initial graph).