Calculated fields are most likely going to be your best friends here. I would separate the start and end times in that field into two separate fields by doing this:
Then you should be able to just graph according to the start or end times, whichever is more applicable. For more "Dateparse" info visit this: Understanding the DATEPARSE Function | Tableau Software
For more string calculations (ie. "left()" and "right") info visit this: String Functions
Actually you will run into an issue with the "00" time at the start. To fix this change the calculated field statements to:
if left([time],2)='00' then dateparse("hhmm",('12'+right(left([time],4),2)))
if left(right([time],4),2)='00' then dateparse("hhmm",('12'+right(right([time],4),2)))
I know its complicated and I tend to overcomplicate things but that is the way I know to deal with "00" values without getting errors.
You need to make sure that you follow the date Timstamp format for ODBC. Refer to this link from Micorsoft on ODBC Datetime Format.
In my case, I already had a date field that was being tabled in Splunk. In Splunk, I cloned existing Time field and changed the format to:
|eval TS=strftime(_time, "%Y-%m-%d %H:%M:%S") TS stands for TimeSpamp.
That did the trick. The new TS field comes into Tableau from Splunk as a date field, not as a string. The order of the YMD matters, as I tried a few other formats.
With this coming into Tableau as a Date Field now, I can do incremental refresh's.
Hope this helps, and will save others countless hours of troubleshooting this issue.