A fat orthogonal search method for biological time-series analysis and system identification

The fast orthogonal search method is illustrated for carrying out both system identification and time-series analysis of biological processes. It is first shown how the method can be used to rapidly obtain concise and accurate difference equation models of nonlinear dynamic systems. Then it is considered how the fast orthogonal algorithm enables accurate identification of cascades of alternating dynamic linear and static nonlinear sub-systems from short data records. Finally, it is illustrated how the method achieves accurate, parsimonious sinusoidal series representations of time-series data. It is shown that the method is capable of precise detection of component frequencies in time-series heavily corrupted with noise, demonstrating finer frequency resolution than a conventional Fourier series analysis.<>