I added a trend line to a 23-month data set and want to continue the trend line out another 5 months. Is there any way to force the trend line to continue into periods where there is no data? This type of very basic forecasting would be helpful! I want to avoid our users exporting data and taking it into Excel.
Hi @Becca_Perry, Thank you for sharing your use case. Currently, we do not support extending trend lines into future periods where there is no data.
However, we found a workaround to achieve it: Create a Date Dimension Model then use it as X-Axis and set the date filter to next x months end. You can check the demo below:
I hope my reply is helpful to you. We also logged your request into our product backlog. We will certainly notify you if there are any updates regarding this feature.
That was helpful, but it’s still not quite s I was expecting, but I think it will be OK for my purpose. I followed the instructions to create a Date Dimension Model (thanks!). I used the date range 2010-01-01 to 2050-01-01.
When I explore the data from just my Date Dim model directly, I can correctly get it to display all the months within a year or all the days within a month.
I added this to a dataset and then tried to display a measure (sum total sales) for every month in the current year. Unfortunately it only shows 9 months January-September instead of all 12. I was expecting to see nulls or 0’s for the future months.
I can get all months to show up if I add another measure: a count of my date dim.
I can also get all months to show up if I select “Show rows with no data,” but we’ve avoided this because we’ve seen it interact oddly with conditions.
I think this will work when I get back into my forecasting use case because then I will have values to display in each month.
Thanks for the help – I would love to see more forecasting-focused features in the future!
Thank you for giving it a try and sharing your experience with us.
Regarding the Show rows with no data feature, we have documentation that explains its behavior in detail. You can find it here.
I noticed you mentioned that you can display all the months by adding another measure, specifically a count of your date dimension. As an alternative, you can also achieve the desired result by moving the count of your date dimension to a filter and setting a condition like “is not null.” This approach will help hide the count from the chart result while still ensuring all months are displayed.
I appreciate your feedback regarding forecasting features, and I will let you know if there are any updates on this front.
Thanks again for your input and patience.