Handling Gap in Data Stream

elif
elif
in Modeling

Hi all,

Due to a 2-3 weeks of gap in the data stream, I need to ignore the data for those weeks from all computations in an Anaplan model that is nearly 50GB in size. If those weeks remain as null/no data, those cells would still remain part of the computation particularly if the logic indicates Sum/#N Weeks or moving average. The data gets loaded automatically into Anaplan from SQL via Informatica. In excel, the solution would be simply not listing/deleting those weeks from the spreadsheet. I've been looking into a solution to handle the gap in the data stream in Anaplan. Recommendations would be appreciated. Thanks!

Answers

  • AjayM
    AjayM

    Hi @elif ,

    Can we not centrally maintain a boolean (in SYS time module?) to identify such ignorable weeks and make this boolean as basis for all the underlying calculations, and also for filtering on UX if you need?

    And, for ‘average’ sort of calculations, you can define a workaround to count # of weeks based on the boolean instead of counting all weeks of the model calendar.

    Cheers!

  • elif
    elif

    Thanks for the reply @AjayM ! Currently, there is no such a boolean to indicate specific weeks that are part of the moving average or any other computations. After creating a boolean, all formulas require an update to refer the boolean line item. Given the model size, I'm concerned about the time and effort required to complete the task. That is definitely a good solution, thanks again for your input. If there is a speedy solution, please share your insight.

  • JMVasicek
    JMVasicek

    Hi @elif,

    Let me make sure I understand the issue:

    • You have a model which gets a data feed where the data lags 2-3 weeks behind the current week
    • This model calculates run rates/moving averages using the last X weeks
    • Because of the above, the data lag throws off the averages

    Is this accurate? If so, is the lagged data structured such that you could derive the number of “missing” weeks and adjust your moving average calculations accordingly? (For example, take the average of 6 weeks ago through 4 weeks ago instead of the last 3 weeks.)

    I apologize if I've misunderstood the issue.

    Thanks!

    Jordan Vasicek

  • elif
    elif

    Hi Jordan,

    Thank you for the reply! You are right, and accurately captured the point. The last for the last 2-3 weeks need to be fully eliminated from the model to avoid any miscalculation. To answer your questions, there is no module with boolean formatted weeks for me to select specific weeks to include or eliminate from calculations.

    Please let me know if you require further information, and thanks again for your time in assisting!

    Elif

Categories