The C3 AI Demand Forecasting application uses a comprehensive data model to represent products, time series data, external factors, and forecasting models.This page outlines the most common terms and C3 AI Types you may encounter while deploying, configuring, and using the application’s data model.
The following terms will be used throughout this guide to describe the C3 Demand Forecasting application.
Item: An entity that is exchanged within a Sales Order, such as a specific good or service.
Sales Order: A transaction through which Item(s) are exchanged for currency.
Customer: The external entity that submits a Sales Order to the business.
Demand Forecast Subject: The Demand Forecast Subject (DFS) is the chosen scope of which forecasting is performed. Typically the DFS is the smallest individual component of a Hierarchy.
Demand Forecast Model: The deep learning model used to generate Demand Forecast Predictions using historical and external data.
Demand Forecast Prediction: A individual datapoint or series of datapoints that represent the Demand Forecast Model’s predition of a future state of a Demand Forecast Subject.
WAPE: Weighted Average Percent Error. A measure of model accuracy (lower is better).
Planning Desk: A scope that a planner (or group of planners) is responsibile. By defining Planning Desks, data access can be limited through Row Level Access Control.
Alert: Demand Forecast Predictions can trigger Alerts if they meet certain user-defined criteria.
Demand Forecasting consists of the following categories of core data types that implementa various functionality of the application.
Entities: These Types define the core of the data model and describe the contents of the data model to be forecasted. Item, SalesOrders, and Customer would fall under this category.
Hierarchies: These Types define the organizational hierarchy of the Entity Types described above and allow for forecasting across various aggregation levels.
Forecast: These types define how Forecasting is performed within the application. Examples within this category are DemandForecastSubject, DemandForecastPrediction, DemandForecastSubjectScore, and DemandForecastModel.
Consensus: These types allow for the importing of external forecasting data series to compare and build consensus between forecasting models.
System Observability: These Types monitor and report on the behaviour of the Application. Examples of these Types include DemandForecastAlertRule, DemandForecastComment, and DemandForecastActivity.
Configuration: While many of the above Types can be modified to configure how the application behaves, Configuration Types allow additional configuration without being connected to the larger Data Model. Visit the Hidden Types Tutorial for more details on how to modify these types to configure the application.
For more information on key types, the data model, and how both influence the behaviour of the application, visit the architecture page. Otherwise, proceed to the developer guide to learn how to configure and deploy an Application.