RPS's TimeSeries Package ======================== rps-help@cs.northwestern.edu The TimeSeries package is an extensible toolkit for time series analysis and prediction. RPS's predserver and pred_reqresp_server use this package to provide predictive models. The package also includes sequential and parallel codes for doing offline evaluation of predictive models. The TimeSeries package includes a subpackage, FracDiff, which implements a fractional arima model. INCLUDE FILES ------------- TimeSeries.h - general include maths.h tools.h util.h - used in many places in RPS LIBRARIES --------- libtimeseries.a BINARIES -------- ts_example test_generic - example codes showing how to use TimeSeries impulseresp - impulse and step responses of a fitted prediction filter crossval_generic - sequential evaluation code crossval_master crossval_slave - parallel evaluation code (needs PVM) evaluate - evaluates the quality of predictions in a testcase perftest - Measure fit and step performance of a model mi gfnn - average mutual information and global false nearest neighbor tools a la Abarbanel's methodology for nonlinear analysis