ElectrifAi
June 23, 2021

Machine Learning Helps Entertainment Industry Predict Demand

Most humans are social creatures and will venture outside their homes to shop, dine, and be entertained. As the pandemic forced people to remain home, they turned to online shopping, meal delivery, and streaming services.

As more and more people begin to venture out in the post-vaccine world, many industries will see an influx of customers—especially the entertainment industry. The joy of experiencing an action movie and hearing others gasp or cheer is thrilling. Zoom calls are a great way to connect with others from a distance but don't give you the same connection.

Just as a roller coaster has its ups and downs, so too will theater attendance. How do companies analyze the new trends and predict attendance in this new world? So much depends on an accurate prediction, such as setting employee schedules and ordering enough supplies.

ElectrifAi’s pre-built machine learning model, Demand Forecasting, is an excellent answer to that problem. With this technology, you can predict the popularity of a show and the number of people likely to attend a specific session.

Business success depends on the number of tickets and concessions sold. Here are some use cases that this machine learning model can solve to increase success:

  • Determine the size of auditorium or venue necessary to hold as many people as are likely to attend.
  • Forecast demand of all the schedulable shows based on which schedule optimization will be used.
  • Demand prediction of a show, in general, depends on several factors:
  • Age of the show
  • Day of the week
  • Time of day
  • Genre (e.g., kids, romance, etc.)
  • Number of nearby venues with the same show
  • Screen on which session is scheduled
  • Film format (2D/3D)
  • Weather

The Demand Forecasting machine learning model can predict attendance to help create a better schedule. The attendance forecasting is divided into two major components: Inter-Day and Intra-Day models.

Inter-Day Model

  • Used to predict day level demand for a show on a given day at a given venue.
  • This prediction is completed in three scenarios depending on the time horizon of the prediction (e.g., current week, next week, and long-term).
  • Different models have been built for different types of shows – movie previews, new movies, and holdover movies.

Intra-Day Model

  • Used to predict show level demand for a show on a given day at a given venue.
  • Determines how day level demand for a show translates to attendance for each show.
  • Demand depends on session time, day of the week, and comparable movie titles, screen type, movie clusters, and other nearby sessions of the same show (Demand Shift).
  • A factor of cluster cannibalization and slot cannibalization is also applied while predicting attendance.

Many data sources and features help the machine learning model make accurate predictions, such as:

  • Auditorium Features
  • Occupancy
  • Cleaning time
  • Show Schedule
  • How often does the show run?
  • Start time
  • End time
  • Advertisement or trailer preview time
  • Show attributes
  • Type (e.g., film, play, concert, etc.)
  • Duration
  • Distributors
  • Percentage of gross ticket sales earned
  • Minimum number of seats and showtimes per contract
  • Screen Info
  • Advertisements
  • Trailer previews
  • Ticket/Retail Transactions
  • Profit contribution
  • Ticket sales
  • Concessions sold
  • GBOR (Gross Box Office Receipts)
  • Weather
  • Will inclement weather affect attendance?

Many types of reports are available to review the data output, such as:

  • Schedule Performance Report
  • Demand Forecasts Report
  • Distributor Reports (e.g., session counts, start times, etc.)

The information included in these reports gives the necessary insights needed to make strategic business decisions. Using these insights, selling the most tickets is much easier to accomplish.

What’s the business impact that can be expected by implementing the Demand Forecasting machine learning model in your company?

  • The most popular movies can be placed in the largest auditorium, thus allowing more tickets to be sold.
  • The more people who attend the movie and are pleased by the experience are likely to refer others.
  • Selling more tickets means more concessions are likely to be sold, increasing overall revenue.
  • Intra-Day means inside or within a day. Entertainment companies can use this to determine how day level demand for a show translates to attendance for each show.

The next step to take is to decide whether to attempt creating your own machine learning model from scratch or partnering with an experienced firm that already has a business-ready model.

Building your own machine learning model is a risky and expensive venue. There are overhead costs when hiring your own team of data scientists and it can take months if not years to produce a working model. This decision is also not guaranteed to produce actionable insights to help your business.

With ElectrifAi, we can help you get quick results that have been proven to help the entertainment industry accurately predict attendance, increase revenue, and decrease risk. Now is the time to use all the new data you’ve been accumulating during these unknown times.

If you would like to see how your data can work for you, contact us today for a custom demo!

Glossary

Cluster Cannibalization

Cluster cannibalization is the amount of demand due to the same cluster or similar shows.

Demand Forecasting

Demand forecasting helps companies predict the amount of demand likely for movies and shows (e.g., plays, concerts, etc.) to help optimize schedules and supplies.

Demand Shift

Demand shift are nearby venues with similar sessions of the same shows.

Inter-Day

Inter-Day helps entertainment companies understand venue demand on a long-term basis (e.g., current week, next week, and longer-term).

Intra-Day

Holdover

Holdover prolongs the engagement of a show, such as extending it for another week.

Slot Cannibalization

Slot cannibalization is the amount of demand due to shows in very close-by slots.