Time series analysis
- Fast Fourier Transforms
Metasite with annotated links to libraries and source
codes,benchmarks, and didactic material on various forms of the Fast
Fourier Transform. From FFTW. These include:
Subroutine library in C for efficient computation of the
Fourier transform in one or more dimension for datasets of arbitrary
size. Winner of Wilkinson Prize for Numerical Software. From
Matteo Frigo and Steven G. Johnson of MIT.
- FXT library
Algorithms fast Fourier and other transforms, sorting and
searching, permutations, and more with associated text book. By Jorg
- Variable star software
A variety of programs for amateur astronomers from the American
Association of Variable Star Observers including: averaging, polynomial
fitting, Fourier analysis, residual analysis (TS); iterative
pre-whitening (CLEANest); epoch folding plots (PhasPlot); and
non-stationary periodicities (WWZ).
periodicities with splines
Nonparametric method for locating periodicities in unevenly
spaced univariate data based on efficient least-squares fitting of
cubic B-splines. Developed for astronomy by C. Akerlof at University of
- Standards Time Series and
Regresion Package (STARPAC)
Library of about 150 Fortran subroutines for time series analysis
and nonlinear regression, developed at the National Institute for
Standards and Technology. Includes digital
filtering, complex demodulation, univariate and bivariate correlation
and spectrum analysis, ARMA and ARIMA models. Old-fashioned
interface. Distributed by the National Center for Atmospheric
Maximum-likelihood autoregressive analysis including
autocorrelation functions, ARMA model fitting, windowning and
filtering, noise and model simulations, spectral density, and
Box-Jenkins analysis. Accompanies volume TIMESLAB:
A Time Series Analysis Laboratory by H. J. Newton (1988).
Distributed by Statlib.
Nonlinear time series analysis package for chaotic systems.
Includes linear tools (autoregressive models, power spectrum, filters),
visualization tools, embedding and Poincare sections, nonlinear
prediction and local polynomial models, dimensionality adn entropy
estimation, Lyapunov exponents, simulations, and cross-correlation of
multivariate time series. From TISEAN Project at MPI Dresden and
University of Wuppertal, Germany.
Metasite with extensive description, software, bibliography, and
Web sites for wavelet methods. Provides linksto nearly
wavelet software packages and toolboxes (many are commercial and not
public domain). Environments include UNIX,
X-Windows, Windows, Macintosh, C, C++, Matlab, Maple, PV-WAVE, Khoros,
Mathematica, S+, IDL and MIDAS. See also a similar site at the University of
- Multi-Taper Spectral
Analysis Methods (MTM)
Spectral analysis with tapering to minimize spectral leakage and
Econometrics package for Windows including: variable
transformations, kernel density estimation, time series analysis
(cross-correlation, stationarity tests, ARIMA & GARCH modeling),
linear regression models (Poisson regression, Tobit, 2-stage least
squares, user-supplied nonlinear), and more by H. Bierens (Penn State).
Package for multivariate nonparametric time series analysis by
maximum likelihood polynomial expansion permitting uneven weighting,
non-normality and nonlinearity. By A. R. Gallant and G. E. Tauchen, and
distributed by Statlib.
Dynamic XWindow graphical system for exploratory time series
analysis including autoregressive models and Fourier analysis.
Distributed by Statlib.
- RQA software
Software for Recurrence Quantification Analysis of complex (e.g.
chaotic, nonlinear and nonstationary) systems. By. C. L. Webber of the
Department of Physiology at Loyola University Chicago.
- Singular Spectrum
Analysis - MultiTaper Method (SSA-MTM)
Toolkit of UNIX utilities to analyze short, noisy time series
testing for trend and oscillating components, and three kinds of power
spectra (windowed correlogram, multi-taper method, maximum-entropy
method). From the Department of Atmospheric & Oceanic
Sciences at University of California Los Angeles.
likelihood ARIMA model
Calculates the maximum likelihood estimators of the parameters of
a fractionally-differences ARIMA(p,d,q) model with covarances and
correlation matrices and standard errors. By C. Fraley of
University of Washington.
Package for optimal and adaptive kernel time-frequency
representations for long signals with changing time-frequency
characteristics. It adapts the kernel over time for improved
performance. By R. Baraniuk (DSP Group, Rice University) and
- ARMA model
Maximum likelihood estimation of autoregressive-moving average
model by Kalman filtering. Applied Statisistics algorithm #154
distributed by Statlib.
- Goodness-of-fit test
for ARMA model
Applied Statistics algorithm #194 distributed by Statlib.
- Change point
for Poisson process
Continuous-time estimation of a change-point in a Poisson
process. By. R. W. West and R. T. Ogden of University of South
- Restricted maximum
likelihood estimation for unevenly spaced data
Parameter estimation for unevenly spaced univariate time series
using a first-order process with white noise. By S. P. Smith of
EA Engineering and distributed by Statlib.
- CUSUM charts
Probability and distribution functions of run lengths in a
cumulative summation chart. By F. F. Gan and distributed by Statlib.