Predata uncovers predictive behavior by applying machine learning techniques to online activity

Predata is the leading predictive analytics company for anticipating geopolitical risk. With alternative data and machine learning, we quantify meaningful online behavior to help anticipate events around the world.

Using a patented methodology, Predata analyzes web traffic metadata to isolate research interest from professionals and academics, distinguishing superficial media coverage from well-informed concern. By selecting audiences and content that reveal a clear intentional narrative, we derive time-series signals to quantify topics and risks that would otherwise defy measurement. As leading indicators of where interest and market activity may be headed, our signals enable customers to focus attention on salient tail risks.

Generally, Predata’s signals represent web traffic as multivariate time series with dimensions for language, browser platform, and time of day. Each dimension, and the interplay between them, is an opportunity to isolate more relevant audiences to your topic of interest.

Fig. 1: Multivariate time series using metadata to compare the English-language interest towards three different collections of online content, each related to different aspects of the Chinese aluminum market.