Most deployments don't back up the operational logs and just accept that in the event of monitoring node failures they just lose the historical logging. Most send the syslogs they want to something like splunk anyways. Losing the operation logs is an admin impact, and not a functional one.
You can filter your TACACS/RADIUS logs in the way you want by setting up a collection filter. You can do this from administration > logging > collection filters.
https://<ise admin ip>/admin/#administration/administration_system/administration_system_logging/collection_filters
TACACS filtering will only work if you are on ISE 2.4p6+ or 2.6+.
https://bst.cloudapps.cisco.com/bugsearch/bug/CSCvb45390