Misleading trend line

ST_Superman
ST_Superman Domo Employee
in Archive

When viewing sales history for a customer and looking at the trend for a period of time, I have noticed that the trend line can be misleading.  The first example is in Domo, you can see that there were no purchases in Feb or Nov of 2017 for this customer. 1.png

 

 

However, the trendline is very flat.  If you were to include 0 for the two missing months, the trend line should look like this:2.png

 

 

I have tried to graph using ifnull(`Net Test QTY`,0) but that did not change the trendline either.  

 

Any suggestions?

 

Comments

  • BlueRooster
    BlueRooster Domo Employee

    What up Scott!

     

    So you basically need some way to add a data point for the missing dates to your data source (Without a date and 0 value the linear regression formula is ran for N-2 essentially)

     

    You could left join a list of calendar dates to your source data and fill in the nulls with 0.

     

    Once you have a reference point for it to assign 0 values to it will look like this:

    image.png

     

    Let me know if you need me to go into further detail. 

     

    Sincerely,
    ValiantSpur

     

    **Please mark "Accept as Solution" if this post solves your problem
    **Say "Thanks" by clicking the "heart" in the post that helped you.

This discussion has been closed.

Categories