I'm analyzing data representing phone calls into a system. I'm trying to come up with a visualization to show what factors contribute to people calling in multiple times.
We key off of the caller's account number (sanitized in the attached workbook and below) to determine whether a call is a 'repeat call' or not.
If the caller calls multiple times, we'd like to be able to show:
1. Attributes of the first call placed by a particular account (Call reason, etc.)
2. Attributes of subsequent calls placed
3. Aggregate on the combinations of things that happened, rather than the things themselves
For example, this is intended to show the distribution of # of calls per account. So most people call 1 or 2 times, some call 3 times, and then outliers are 4 or 5 times.
I'd like to take the next step and be able to produce calculations based on those bins of # of calls.
So for example, for people calling in 2 times, what was the distribution of their call reasons? What was the most likely call reason on the 1st call to trigger the 2nd call? And so on.
My data looks like this:
I think what I need to start would be calculations showing:
1. For each account #, how many times did that person call
2. For each call, which # call was it: 1st, 2nd, etc.
I can do #1 easily directly in Excel. The problem with embedding it into the source data is I end up with too many records. For example, if an account called 5 times, then for each record in the data, there will be a 5 in that column. So 5 5's. Which means I can't then aggregate on that column. I think a better structure would be a mapping of account numbers to # of calls.
So instead of this (simplified):
Account Number Call # # of calls
12345 1 5
12345 2 5
12345 3 5
12345 4 5
12345 5 5
Something like this:
Account Number Call #
Account Number # of calls
Is this something that could be represented in Tableau directly?
I know this is a mish-mash of thoughts. If anyone wants to dig into my data a bit and play around to see what might be possible, I'd be very grateful for the help.