roahd 1.4.3

New feature

Added tools for manipulating and visualizing depthgrams (#1, @aefdz). This mathematical constructs aim at facilitating the visualization of outliers in high dimensional functional data sets. The depthgram() function computes a number of depthgrams from the functional data set. An S3 specialized method for plot() makes it possible to visualize the depthgrams and proceed with a visual inspection at outliers.

Improved unit testing: Refactoring of unit tests using testthat 3e edition and in particular snapshot tests for plots and complex objects.
Added an hexsticker.

Switch from Travis to Github Actions for continuous integration.
Setup automatic R CMD check on Windows, macOS and Linux for both the latest release and the development version of R.
Setup automatic deployment of a website for the package that references a package introduction, its help and all vignettes.
Setup automatic computation of test coverage and report to both the Github page and website of the package.
Added CRAN status badge to README.
New package maintainer.

Updated all matrix class checks for compliance with R-4.0 in which the matrix class inherits from the array class.

Fixed dependency error on a new version of scales package that breaks the use of multivariate fbplot in the corner-case of zero outliers.

Extended Spearman’s correlation coefficient computation for multivariate datasets with more than two components.
Added bootstrap-based computation of Spearman’s correlation coefficient bias and standard deviation.
Added methods to provide bootstrap-based confidence intervals on Spearman’s coefficients for two univariate functional datasets or a multivariate functional dataset.
Added a bootstrap-based test on Spearman’s correlation coefficient for two multivariate functional datasets.
Added an outliergram version (without graphical display of original data) of multivariate functional datasets.
Added example multivariate functional datasets of ECG signals.

Added two convenience functions to append compatible functional datasets (univariate or multivariate).
Added a [ operator overload for multivariate functional dataset representation mfData.

Fixed bug in cor_spearman() function. Now the standard Spearman correlation is not computed on ranks of MHI/MEI, but on MHI/MEI itself. The difference is very small, but allows for full reproducibility of the results in the original paper.

Fixed typos in doc.
Standardized formulas for the application of F inflations in outliergram and boxplot.

Removed check for uniformity in the grid of fData() and mfData() constructor.
Added the possibility to subset fData in time with logical vectors.
Fixes in methods BD, BD_relative, HI and EI: the previous computational technique was based on arguments from the popular reference “Exact fast computation of band depth for large functional datasets: How quickly can one million curves be ranked?” by Sun, Genton and Nychka, which in the case of BD, and HI/EI are wrong. Now the implementation exploited sticks to the definition, at the cost of a higher computational burden (and thus, time to complete the computation).

Modified the check of the grid provided to build fData objects. Since support is provided only for evenly spaced grids, a check is needed before building an fData object. Before it was:

all(abs(diff(unique(diff(grid)))) < 1e-14)

Now it is:

max(diff(unique(diff(grid)))) / diff(range(grid)) < 1e-13

which is much more robust in practical cases.

Extended README.md.
Added cov_fun method to compute covariance and cross-covariance functions for either univariate or multivariate functional data. Implemented the S3 class Cov and plotting specialization plot.Cov, wrapping graphics::image.

Fixed typos in documentation.
Fixed typos in vignette.
Added Travis and Codecov support.
Modified the default parameter value for trial_size in fbplot from Data$N to 8 * Data$N.
Added check to fbplot and outliergram that raises warnings when parameters different than those supported are provided through adjust argument.