Forum Discussion

MichaelJackpbi's avatar
MichaelJackpbi
Icon for Helper I rankHelper I
9 years ago
Solved

Get data for 1 week before depend on TODAY()

Hi friends, I would to load data for 1 week only depend on current date!. I.e. Today is 12/14/2014 then my dataset will get data from 12/7/2016 to 12/14/2016 AND tomorrow, my dataset will load data ...
Need Help
  • tringuyenminh92's avatar
    tringuyenminh92
    9 years ago

     

    Hi MichaelJackpbi,

     

    In Query Editor, Choose your date column -> click filter icon in top right of header of column -> choose Date Filters -> Custom Filter -> Fill in This quarter and Last quarter with OR operator as below pictures:

    (Every actions have been done by UI - love it )

     

     

    Back to your speed question, depends on your query(or amount of data) the loading time will be fast or slow. That means it will take more time when loading data for quarter than week. And i hope you have good index stategy in your sources. it's one important factor.

    • Import mode: loading all data of your query - one time when refreshing - good for UX
    • Direct Query: loading data based on user's interaction (user could choose an option in slicer -> system generates the query corresponding and just loads the data of that query) - user has to wait when loading huge data of that action or long distance of data center to user's location

     

     

     

     

GilbertQ's avatar
GilbertQ
Icon for Super User rankSuper User
9 years ago

Hi MichaelJackpbi, it all depends on where your source data is coming from?

 

If it is a database you can do it as part of your source query.

 

If the data is coming from a file, then yes you will need to bring all the data in first and then filter it afterwards on your date column. You can do this in a few ways, but we would need more details in terms of how your data looks to place the dynamic filter. 

MichaelJackpbi's avatar
MichaelJackpbi
Icon for Helper I rankHelper I
9 years ago

GilbertQ THank for fast response!
Ya, i have a database contains few column like ID, Product,Code, OrderDAte,Qty....

I would create a dataset to only update for 1 week before from Now! i means today and can be able to refresh data everyday as following that rule.

 My database is oracle 12C, i try to wirte the query to filter conditions in custom window but i see the speed to load data very slow than using choose tables

Regards,
J.

  • GilbertQ's avatar
    GilbertQ
    Icon for Super User rankSuper User
    9 years ago

    Hi MichaelJackpbi, I would suggest doing it at the source, in doing so if your data is large your Oracle database is much more efficient at only giving you the data you require.

     

    Also then you would have to load all the data into Power BI and then filter it out, which will take longer to load the data.

    • MichaelJackpbi's avatar
      MichaelJackpbi
      Icon for Helper I rankHelper I
      9 years ago

      GilbertQ you mean i have load all of data from database to power bi then filter as my rules! there are 65m rows in database and i just need to get 1 week only- around 2-3m..

      • GilbertQ's avatar
        GilbertQ
        Icon for Super User rankSuper User
        9 years ago

        Hi MichaelJackpbi, yes that would be correct.

         

        The reason is that if your query from Oracle, said give me all the data, it would then give you the 65m rows. Which would then be loaded into Power BI. And then you would then apply the filter to filter out the rows to leave you with 2-3m rows.

         

        So it would be best to write the Oracle TSQL to only get the last weeks worth of data. 

         

        Something like this after doing some Googling

         

        where adddate >= next_day(trunc(sysdate), 'MONDAY') - 14 and
      adddate < next_day(trunc(sysdate), 'MONDAY') - 7