Audience Forecasting for Digital Video Content

In a digital medium environment to generate a forecast value, a method implemented by at least one computing device, the method comprising:

• receiving, by the at least one computing device, a request identifying an audience segment and a video series for which the forecast value is to be generated;
• obtaining, by the at least one computing device, data describing past audience impressions for the audience segment of past episodes of the video series and an availability schedule specifying a time at which the past episodes are made available for consumption by the audience segment and a time at which a future episode of the video series is to be made available for consumption by the audience segment;
• determining, by the at least one computing device, a weighted regression coefficient describing diminishing daily lag effects of the availability schedule on the past audience impressions over time;
• generating, by the at least one computing device, a linear regression model based on the past audience impressions, the weighted regression coefficient, and the availability schedule;
• generating, by the at least one computing device, the forecast value for the future episode based on the linear regression model, the forecast value describing future audience impressions for the future episode based on the linear regression model as a linear function of the past audience impressions and the availability schedule;
• and
• outputting, by the at least one computing device and based on the forecast value, a selection of digital marketing content in response to the request.

In a digital medium environment to generate a forecast value, a system comprising:

• means for receiving a request identifying an audience segment and a video series for which the forecast value is to be generated;
• means for obtaining data describing past audience impressions for the audience segment of past episodes of the video series and an availability schedule specifying a time at which the past episodes are made available for consumption by the audience segment and a time at which a future episode of the video series is to be made available for consumption by the audience segment;
• means for determining a weighted regression coefficient describing diminishing daily lag effects of the availability schedule on the past audience impressions over time;
• means for regressing the past audience impressions and the availability schedule to generate a linear regression model based on the weighted regression coefficient;
• means for generating the forecast value for the future episode based on the linear regression model, the forecast value describing future audience impressions for the future episode as a linear function of the past audience impressions and the availability schedule;
• and
• means for outputting the forecast value for presentation at a user interface in response to the request.