5 Replies Latest reply on Jun 24, 2018 3:52 PM by Chris McClellan

    Incorrect results in Tableau 2018.1 extract data source




      I'm using a view from a MySQL database containing about 150 million rows as my data source. Because of the size of the number of rows and the fact that the data will never be updated, I'd like to create an extract in Tableau 2018.1 to improve performance. In the extract, I've selected all rows and the aggregate data for visible dimensions option, and I have set my default aggregation for all measures of interest to average (the data do not have a time dimension). When I create the extract, I am warned the extracted data may cause the views to be different than they are under the live connection because I'm using the average for aggregation and have a few calculated fields and one bin dimension.


      I'm not getting the expected results when I use the extract. If I run the extract and live data sources for a smaller subset of the data, I do not get the same results. If I write a query to obtain equivalent aggregations directly in SQL, I get the same result as given by the live data and not the extract. Interestingly, the incorrect values only appear to affect the values of one measure and not the others (I've not compared the exact numbers, but they make sense). I've tried deleting the extract file and starting over several times, but this doesn't work. I've also tried recreating the view in the SQL database, but this doesn't help, either.


      Am I having issues related to the warning I was presented when creating my extract? If so, is there anyway to create an aggregated extract using the average without issue?


      Any help would be tremendously appreciated.


      Have a nice day!