Categories
- All Categories
- 15 Oracle Analytics Sharing Center
- 14 Oracle Analytics Lounge
- 214 Oracle Analytics News
- 42 Oracle Analytics Videos
- 15.7K Oracle Analytics Forums
- 6.1K Oracle Analytics Idea Labs
- Oracle Analytics User Groups
- 78 Oracle Analytics Trainings
- 14 Oracle Analytics Data Visualizations Challenge
- Find Partners
- For Partners
OBIEE 12c 12.2.1.2.0 - How to get forecast function working for small time intervals?

Hello everyone,
i'd like to know whether it is possible to debug the forecast function, included in Oracle 12c as forecast seems to be incorrect, if the forecast is based on just a few Input values (e.g. < 10 time values ).
Is there a possibility to get forecast working for time intervals with less than 10 time periods (years, months etc.)?
Background Information/Example:
I built an simple test answer, trying to forecast business revenues for 3 years ahead, based on the last 5 business years.
Forecast results seemed fine, when the forecast was based on a lot of input values (e.g. more than ten years). But when there were less input values (e.g. less than 10 years), then forecast-values stayed the same for every year.
OBIEE 12c Version used: 12.2.1.2.0 (on premise)
Simple test answer:
The answer includes 3 columns: Time.Year, Facts.Revenue, Forecast
My column formular for the forecast column looks like this:
Forecast column formular |
---|
FORECAST("Facts"."Revenue", ("Time"."Year" timeYear),'forecast','modelType=arima;numPeriods=3;') |
This is what the results looked like, when there was no filter on the year column. Results seemed to be ok/ as expected.
Year | Revenue | Forecast |
---|---|---|
2005 | 100.000 | |
2006 | 110.000 | |
2007 | 120.000 | |
2008 | 130.000 | |
2009 | 140.000 | |
2010 | 150.000 | |
2011 | 160.000 | |
2012 | 170.000 | |
2013 | 180.000 | |
2014 | 190.000 | |
2015 | 200.000 | |
2016 | 210.000 | |
2017 | 220.000 | |
2018 | 230.000 | |
2019 | 240.000 |
Decreasing the time interval by putting a filter on the year column (e.g. YEAR >= 2012) instead, led to wrong results. Forecast was the same for the years 2017, 2018, 2019.
220.000
Year | Revenue | Forecast |
---|---|---|
2012 | 170.000 | |
2013 | 180.000 | |
2014 | 190.000 | |
2015 | 200.000 | |
2016 | 210.000 | |
2017 | 220.000 | |
2018 | 220.000 | |
2019 | 220.000 |
Nachricht geändert durch 1454010
Message edited