Please check the below,it should help you out hopefully.
- Incremental Load with Tableau - YouTube
- Re: Is it possible to delete certain data from Extract?
- Re: difference between incremental refresh and full refresh in tableau extracts
- incremental refresh | Drawing with Numbers
hope it helps you.
I do understand the difference between incremental and full, however my issue I believe it is different. I am using a combination of both (incremental for daily refreshes / full at the end of the week) but I would like to limit the full extract to only last week of data without deleting all the data which was already there, although it seems it may not be possible ..
It is possible to selective update/refresh extract using WDC and date and time stamp fields,please check few additional links:
DB filters can be applied too.
DB filters and limit function
hope it helps.
2 of 2 people found this helpful
If I'm understanding correctly the data source can change for 2-3 days, then will be set? There is a work around I've used, which will work if you can use blended data sources for the view.
You will need to set up one data source for the historic data with a incremental refresh that has a filter so that it only loads data from the point where the data won't change, for example three days ago. This can always run as an incremental source if the data is not going to change.
You will then need to set up a source for the last 3 days which runs as a full refresh so that it is always up to date and is only pulling the latest data.
Thirdly you will need to create a scaffold source which has the dates but no data behind it which you can use to link the tables. This link should describe this part of the process.
Yes, this is exactly the case. However as I wasn't sure exactly how long the data could change I was saying one week to be safe. Your solution is exactly what I was trying to do, I just wasn't able to blend the sources since they were coming from the same source. I will review the "Data Scaffolding" trick and try it with my data.
Great. You need the scaffolding to create the dates behind the data, otherwise you won't be able to blend due to the fact the dates won't match across the two data sets. Once that's in place it should hopefully be fairly straightforward.