STSA - The Statistical Time Series Analysis Toolbox for O-Matrix
Version Enhancements and Additions Summary
The STSA (Statistical Time Series Analysis) Toolbox Version 2.1 release
includes a large number of enhancements and additional functions.
Some of the new areas of functionality include:
Some specific version 2.1 enhancements include:
- There are 2 new sub-directories, or functional categories
that expand the existing capabilities of STSA:
POD; Proper Orthogonal Decomposition and Singular Spectrum Analysis, and
NONPAR; functions for nonparametric, nonlinear time series analysis.
- All sub-directories have been updated, enhanced, and expanded with
new functions - STSA can now handle a greater array of time series problems.
- More functions have been added that can be used and in
non-time series contexts: more random number generators, cumulative distribution
functions, probability density functions, statistical tools, and
- Functions for performing singular spectrum analysis (SSA) of a
time series including decomposition,
reconstruction and forecasting.
- Functions for handling nonlinear time series using
including local polynomials, cubic splines, functional
coefficient models, partially linear models and various cross
validation methods for automated bandwidth selection.
- Enhanced statistical tools (logistic regression for handling
binary time series, enhanced QQ plot function).
- More examples that illustrate and expand
on existing and new functional capabilities.
- New samples using provide real-world, and simulated data sets.
The STSA Toolbox Version 2.0 release
included a large number of enhancements and additional functionality
Some specific version 2.0 enhancements included:
- Four new sub-directories (FILTER, OPTIMIZE, RNG & STATS) that expand
the capabilities of STSA.
- All sub-directories have been updated and expanded with new functions -
STSA can now handle a greater array of time series problems.
- STSA now contains additional functions that can be used and in non-time series
contexts (random numbers, statistical tools, generic optimization).
- Many functions now contain formatted screen output that greatly enhances the
speed and quality of any analysis.
- Compute the theoretical autocovariances of an ARMA model.
- Durbin-Levinson-Whittle algorithm for computing innovations.
- Perform Granger causality tests.
- Compare forecasting performance of competing models.
- Compute robust estimates using Least Absolute Deviations.
- Filter a time series using a variety of filtering methods and models
(Savitzky- Golay, generic finite impulse response, time-invariant Kalman filter
with estimation, Holt-Winters with seasonal and estimation).
- Estimate the empirical probability density and cumulative density functions.
- Bootstrap a time series using the maximum entropy bootstrap.
- Estimate ARMA-GARCH models.
- Enhanced nonlinear optimization functions with optional screen output.
- Random numbers from various statistical distributions.
- Enhanced frequency domain functions (additional methods for estimating the spectrum,
estimation of cross-spectrum, squared coherence, amplitude and phase, additional
methods for long-memory models, enhanced plots).
- Enhanced statistical tools (regression with full screen output,
enhanced plots, transformations and tests for Gaussianity, PCA and Factor Analysis).