PlanIQ Backtest data automation and Backtest data should come for all Actual periods as we import data let's suppose of 60 months for forecasting of 3 months then Backtest result come out for only past 3 months not for 60 months
We cannot produce backtest for all the history because it works differently.
Let's assume we have 24 months of historical data and we want to forecast 6 months ahead.
When algorithms are training, PlanIQ will take the 24 months of history and split it into two parts. First 18 months are being used for training. The latter 6 months are being withheld from training process and during the model creation, PlanIQ is actually "forecasting" this period. The "forecast" produced is then compared to 6 months of actuals that were kept on the side. This is how quality metrics are generated.
As a result, we cannot provide backtest information for the entire history.
However, if you write backtest results into the same module where you will be writing your forecasted values, you will be able to continuously track performance of PlanIQ forecast compared to your actuals - as you are adding more actuals and generating more forecasts.
Hope this helps.
As an Anaplan Tenant Administrator, I need the Export functionality to be able to export all the existing Tenant Level Roles information as well as the role details, into a CSV file, So that I have the accessibility to document and report to my security team. I know this feature is successfully implemented when I can see…
Currently, only Workspace Admins in Anaplan can access certain model metadata and deployment validation panels (such as Revision Tags, Data Tags, etc.) in the Production workspace. For security and compliance reasons, we want to avoid granting admin-level access to users who only need to view this information. There is no…
Currently, Anaplan does not natively restrict number inputs to specific ranges. Users can input any numeric value, which may result in data integrity problems, especially in collaborative or decentralised planning environments. Introducing a validation capability—where model builders set upper and lower bounds for number…