Historical data is mandatory for PlanIQ forecasts. Related data and attributes are both optional. Historical data is sometimes called actuals or historical actuals.
Here's what we recommend you get ready before you start to use PlanIQ:
Source modules for your data collection setup
Create modules for the different datasets you want to export to PlanIQ for your data collection setup. You can choose to exclude some of the values from your forecast.
Note: Make sure you set your model calendar settings to cover the length of your forecast horizon.
Historical data module (required)
Your forecast requires historical data to base the predictions on. We advise that you build a module that contains a time dimension and the essential data that you need predictions for. Then you can export the module to PlanIQ with an export action.
Related data module (optional)
If you want to include a related data set to influence your forecasts, we recommend you create a module in your source model to hold the related data. Related data can enhance the accuracy of your forecasts.
Examples of related data are forward or historical looking marketing promotions.
Include a time dimension and a list of IDs for historical and related datasets.
Attributes module (optional)
You can also add a module for attributes. Attributes group items by their shared characteristics. For example size, or color. This is particularly useful for forecasts based on neural network algorithms, such as CNN-QR or DeepAR+, as they improve the results of forecasts where some items are new. That is, items that have been newly introduced to the forecast.
Ensure you configure an export action for each source module you create so you can export the data out to PlanIQ.
If you want to exclude certain values from your forecast, you can configure this in your source data.
Prepare a module to import your forecast results
Once a forecast action generates your predictions, it needs a module to import and save those predictions into.
Forecast results module
Create a module that forecast actions can import your predictions into. You can use this module to store and compare different forecast results.
Configure an import action to import your predictions from PlanIQ to your forecast results module.