• I think I have it sussed, I finally got it to return *exactly* the same result as Excel. I think the problem (re: the last param inclusion or noninclusion) relates to one little line in the BOL

    quote:


    Note Empty cells or cells containing text or logical values are ignored; however, cells with values of zero are included.


    . So, in my testing I was using a time dimension (foodmart sales) as suggested is a good option for leaving out the last param BUT because the time dimension is returning strings (not numbers), these values are essentially being ignored for use as the x-values. SO to get the same result as the excel 'TREND' fn, I created a simple calc member that interrogated the [Time].CurrentMember, looked to see if it was Month 1, 2 or 3 in Q1 of 1997, if it was, then the measure value was a 1, 2 or 3. I then used this measure as the last param in the function, and sure enough, period/month 4 was predicted and displayed the same value as Excel's prediction.

    I can send through the MDX if this will help. So, I think it is related to the exclusion of text items as valid values for the last param. This is really a flaw in the functions methinks! Should possibly be added to the Newsgroups as a Qn to MSft.

    Steve

    Steve.