Joshua Garland

In a new paper for Physical Review E, SFI External Professor Liz Bradley and colleagues Joshua Garland and Ryan James quantify predictability, with a strategy for determining which predictive method best suits a given system. Bradley says the paper was "nucleated" at SFI. 

The authors use techniques from information theory to construct a strategy for evaluating whether a given time-series prediction method fits a given set of data, and demonstrate their strategy on a variety of synthetic and real-world data sets, from lasers to the signals inside microprocessor chips.

Read the SFI working paper (May 14, 2014)

Read the paper in Physical Review E (November 12, 2014) 

More SFI News