After you set up your data collection, the next stage is to create a forecast model. PlanIQ trains the forecast model so it can identify the data patterns that are in the data collection. Once trained, the forecast model uses machine learning and statistical algorithms to generate the predictions.
Your feedback is important to us. To ensure you have a good content experience, we'd appreciate if you could tell us how you find the guidance on this page and if it addresses your questions.
To take part, complete the quick survey at the end of the article, and tell us how we can improve.
Ensure you have one or more data collections set up successfully.
To create your forecast model:
- Click New forecast model on the Forecast models page.
- Enter a name for your forecast model.
- Select a data collection. If you do not have one ready, create one.
The data collection trains the algorithms so they make predictions based on the patterns in the historical data.
- Select the algorithm based on your data set in the data collection or keep the default selection.
The default, Anaplan AutoML, chooses the best algorithm for your data set. PlanIQ works out the maximum length and number of intervals you can select for your forecast horizon. This is based on the information in the data collection.
- Select the time interval and how many intervals you want to see in your forecast.
- Optional: Select a country from the Country-specific holiday calendar dropdown.
This is a good option if you want to take into account a country's public holidays. Whichever country you select, their public holidays are included as an additional related data set in your forecast.
Tip: Calendars are useful for forecast models that are based on data from a single region. To use this option, ensure the algorithm supports related data sets.
- Click Create forecast model.
We recommend you check back later and give some time for the model to build. When the model is created, a notification displays in the top-right.
The right pane displays an overview of the model properties and metrics.