Skip to main content
Documentation
Documentation
Administrator Guide
Last Updated: 2026-02-06
Concept: Scheduling Events

Concept: Scheduling Events

Scheduling events are user-designated dates or periods that signal to the machine learning (ML) model that the observed performance of a business metric on that day is significantly atypical due to a specific, unique, or recurring circumstance (e.g., a major holiday or an unforeseen one-time event). By tagging these dates, you can instruct the model to isolate the event's influence, heavily weighting the relationship between these dates to the exclusion of the normal trend. This makes the tag a powerful tool for modeling abnormal metric behavior but less effective if the user intends to maintain the integrity of the long-term forecast for normal business operations.

Scheduling Event Calendar

A scheduling event calendar is a collection or grouping of individual scheduling events used to provide specific contextual information to machine learning (ML) forecasting models.
This calendar is defined and maintained for one or more scheduling organizations, allowing users to target distinct sets of atypical dates—such as holidays, promotions, or one-time incidents—to different business segments. By applying tailored calendars, the ML models can more accurately align metric forecasts to the unique performance patterns influenced by geography, market segment, or other operational differentiators. Example: Separating events for
APAC Retail
 from
North America E-commerce
.

Reoccurring Event Tag

A Reoccurring Event Tag is a reusable label that you can apply to multiple scheduling events (both past and future) to inform the machine learning (ML) model that the tagged dates are related or similar in their impact on business metric performance. By linking these dates across time, the ML model can analyze the collective historical activity associated with that specific type of event. Example: Tagging New Year's Day across consecutive years. This process allows the model to appropriately weight and anticipate the atypical metric performance when that specific reoccurring event occurs again in the future.