Moving Average without dates

Phil, thank you so much!

Hello Phil_Seamark once again :)

I am still stuck with the task. I need a average for prices that are collected on a weekly basis, but often the days are omitted, for that I have created an index that reset after change. It has 4 levels - Region, Municipality, Product Category, Product Name - after we move to next municipality with different category and product, index is reset.

When I use your formula on this file, I get huge numbers, I think it is summing everything beneath without taking into consideration index.

Thanks in advance!

Best,

Levan

Phil_Seamark you, sir, are a HERO.

I was trying to create moving averages using dates, but every time I filtered the date to show only the last 12 months (and I have 5 years of data, so it needs filtering) the measure would crash. Using an index solved the problem, I would never think of this.

Thanks for sharing.

Hi!

Very useful post but I have another question. How can I calculate moving average without creating calculated columns and using only measures? I'm connected to SQL analysis server and I cannot create calculated columns. 

I'd like to calculate average duration of the last 3 Name_IDs.  My goal is to plot Name_IDs on x axis and Moving average as value. 

Input:

 My current attempt:

Moving average = 

VAR index = RANKX (
    ALLSELECTED ( NAME ),
    CALCULATE(MAX(Table1[EndDate])), 
    ,
    ASC
)

Return
    AVERAGEX(
        FILTER(
            Table1,
            indexName >index-3  && -- indexName is another measure that is equal to VAR index
            indexName <= index

           ),Table1[Duration]
           )

 Thank you!

Thanks for the suggestion! How do I resolve the same question, using moving standart deviation? I tried to substitute AVERAGEX for STDEVX.P, but the result was not correct.