Prediction Engines

Опубликовано 8 августа 2016, 18:55

From crowdsourcing to data-driven models, technology offers new ways to collect and aggregate information on an unprecedented scale, allowing researchers to make reliable predictions about elections, policy, corporate decisions, economics, finance, sports, and entertainment. This session will report on progress in the science and engineering of polls, prediction markets, and forecast models. Examples include a fundamental model that correctly predicted 49 of 50 states nine months before the US Presidential election, and combinatorial prediction markets capable of estimating billions of complex predictions like correlations between states. The session will include talks from academic experts in the field as well as Microsoft researchers investigating prediction markets, scoring rules, polls, and forecast models.