On the possibility of predicting ratings reliably

Company: Nielsen

Author: Ian Garland


The paper reports on a secondary analysis of Australian People Meter data, using trend-series data analysis techniques. It places particular emphasis on deriving a neural network model of television audience estimates. The research program has been motivated by two major initiatives: the possibility of deriving a parsimonious function to describe a historic ratings data base and the possibility of predicting audience levels based on the derived function.

  • PDF
  • This could also be of interest