Machine learning augmentation reduces prediction error in collective forecasting: development and validation across prediction markets with application to COVID events.

Where the human-machine hybrid and the human only forecast model disagreed by 5% or more with regard to event likelihood, the hybrid forecast accuracy advantage was substantial, corresponding to highly significant 13% and 14% gains in AUC in the out-of-sample data sets of Almanis B and NGS2, respectively.

also in 132,142 other papers