Improvements to our predicted attendance logic unlocks better data coverage

Published on April 11, 2022
Valerie Williams
Senior Content Marketing Manager

Predicted attendance is a critical data point that powers a variety of event-centric use cases including demand forecasting, surge alerts, screen placement, and many more. That’s why once we aggregate events from a variety of sources, we take events with flawed, missing data and enrich them using models that rank and predict how many people are attending events.

PredictHQ’s predicted attendance is a complex data science process which uses a host of machine learning models which include regression and natural language processing to analyze event titles, descriptions, and times to provide the most accurate representation of an event, including the size. 

Attendance and sometimes venue information is commonly missing, even from our pool of 350+ sources. Plus, even if you have venue information, venue capacity does not equal attendance - the unique composition of each event, from the performer, to the sports team, to even the central topic of the event may attract fewer or more attendees in the same venue. Most event data sources estimate attendance based on venue capacity, which is often inaccurate and results in over or under forecasting. 

With PredictHQ’s extensive knowledge graph, we predict attendance based on many inputs including venue capacity and performer popularity, resulting in a reliable number that you can use in your models. We provide predicted attendance for both historical events and future events, which can be used along with Rankings to understand an event’s potential impact on your business.

This new release further improves our unique ability to rank events and accurately predict attendance– which no other data source provides. Enjoy better coverage for our data and better underlying quality of PHQ predicted attendance outputs, especially for larger volume events.

This improvement models the predicted attendance for events by making use of features including:

  • Concerts

    • Genre

    • Performer average ticket sales

    • Music label (publisher)

  • Sports

    • Teams

    • Match position in the season (qualifiers, playoffs, finals, etc. will have different impact on popularity)

    • Home vs. away game

  • Population density

  • Global factors (such as COVID)

  • Providers

  • Geolocation

  • Categories

  • Labels 

  • and more

Questions about our predicted attendance coverage? Contact our team today.