Forecast line missing at granular level

[PowerDonkey]

Hi All,

Got an odd one for you.  I have a line chart where I want to enable forecasting on.  I have a date field (order date) and calculated column (salesprice - discount)  which shows a line chart, so no problems there.  When I enable the forecast option, I get no line even though I have configured it for 1 year 

Date range is 2016-01-01 to 2017-10-01

If however, i change the order date to a hierachy, it then starts working:
Date hierarchy at year level

Why is it doing this?  I have even used a date slicer to reduce the date range to 3 months and change forcasting to 3 days, still does not appear.

The value is a sum and the order date is defined as date in modeller.

Any ideas from the uber minds of Power BI?

Thanks,

PD.

[v-huizhn-msft]

Hi @PowerDonkey,

Have you resolved your issue? If you have, please mark the right or helpful reply as answer, so more people will benefit from here.

Best Regards,
Angelia

[PowerDonkey]

Hi,

Sorry for delay, other commitments stopped me from responding sooner. 

If I understand you correctly, you are saying every day of the year must be populated a sales value otherwise forecasting will not work?  And because there are values at a month/period/year level (aggregated of all days within that grouping), it does work?

If so, then how do I populate the chart with zeros on the days no sales are performed?  This would be very problematic with all the other dimensions used such as product groups, sales channels etc, that would insert considerable amount of records to backfill the dataset. On a side note, I am having a this issue with cumalitive charts where a product group may not be used one month but is the next, this does not provide carried forward amounts for that product group for the missing month. The only answer I have found is to populate these gaps the same way by adding zero records for that period.

I would have thought the date dimension related to my fact table which is being used as the x axis would by default zero out any missing values on those days?  

Look forward to your response.

PD.

[v-huizhn-msft · Accepted solution]

Hi @PowerDonkey,

>>If I understand you correctly, you are saying every day of the year must be populated a sales value otherwise forecasting will not work?  And because there are values at a month/period/year level (aggregated of all days within that grouping), it does work?

 
If you want to have the forecast line, the date on X-axis must be regular not only every day. It may be every two days. Otherwise, it will display the following error message after test.



>>I would have thought the date dimension related to my fact table which is being used as the x axis would by default zero out any missing values on those days?  

Yes, you can create a calendar table, create a relationship between calendar and your fact table. Then select the calendar[date] as x-axis, it would be zero by default for missing values.

Best Regards,
Angelia

[v-huizhn-msft]

Hi @PowerDonkey,

Your data is continuous, right? If it is, I try to reproduce your scenario, while I get expected result.

Date range from 2013/1/1 to 2015/7/20

Date hierarchy at year level

What's your Power BI version? Please update it to the latest one. And do you mind share your .pbix file for further analysis? You can send it by private message if your data is confidential.

Best Regards,
Angelia