Systems and methods for predicting television viewership patterns for advanced consumer segments

Abstract

Systems and methods create new viewership estimates to drive linear ad schedule optimizations. The systems use machine learning techniques to predict granular-level television viewership metrics for any consumer segment measurable at a national level. The systems provide capabilities beyond merely estimating and forecasting television viewership for age- and gender-based demographic segments. The systems accurately estimate viewership for any consumer segment, including behavioral, demographic, and other segmentation techniques and provide reliable viewership predictions. The systems model viewership by training an ensemble of machine learning models using historical consumer segments data and TV viewership data. The models work in concert to create viewership predictions, which are ingested, transformed, and processed and then used for determining and pricing advertising sales based on predicted viewership for advanced segments, for pacing forecasts and pre-actuals in a campaign stewardship program, and for further model training in an optimization engine.