NEWS.md

In jtrinka/Rfssa_dev: Functional Singular Spectrum Analysis

Rfssa 1.0.0

Updates

• fssa updated to allow for the decomposition stage of multivariate functional singular spectrum analysis (mfssa)

• freconstruct updated to allow for the reconstruction stage of mfssa

• plot.fssa was updated to allow the user to plot:

  1. left singular functions in curves (lcurves)
  2. left singular functions in heat map (lheats)
  3. periodograms (periodogram)
  4. right singular vectors (vectors)

• plot.fssa was also updated to remove plotting options:

  1. meanvectors and meanpaired removed in lieu of vectors and paired plot options
  2. efunctions and efunctions2 removed in lieu of lcurves and lheats

• wplot was updated to allow the user to specify the cuts parameter to improve visualization of weighted correlation (w-correlation) matrix

New Additions

• The functional time series (fts) class was added which extends functional data (fd) objects

• fts arithmetic was added allowing the user to perform addition, subtraction, and multiplication of fts objects with other fts objects or fts objects with scalars

• Indexing of fts objects is allowed

• Correlation of fts objects can now be measured using cor.fts.

• plot.fts, which uses plotly, was added to allow for visualization of fts objects

• launchApp was added which allows the user to launch a shiny app for univariate functional singular spectrum analysis of multivariate functional singular spectrum analysis based on type specified. This app helps with visualization, and general exploration of the algorithm.

Minor improvements and bug fixes

• Jambi data set was added to create bivariate mfssa example

• Plot ordering in plot.fssa was fixed for types lcurves, lheats, and vectors

• More clear error checking was added to code

jtrinka/Rfssa_dev documentation built on Dec. 21, 2021, 4:14 a.m.