regr_cc_sof()is now set to
fdapackage now can be used also with
funcharts, which previously it could be used only with B-spline basis. In particular, Fourier, exponential, monomial, polygonal, power and constant basis function systems are available.
get_outliers_mfd()allows to find outliers among multivariate functional data using the functional boxplot through the
fbplot()function of the
control_charts_sof_real_time()have been deprecated. Instead, use
regr_cc_sof_real_time(), respectively, with argument
include_covariates = TRUE. This has been done to make more consistent the regression control chart functions for the scalar (
regr_cc_sof_real_time()) and functional (
regr_cc_fof_real_time()) response cases.
alphaparameter in all control charting functions, which previously could only be a list with manually specified values of the type-I error probability in each control chart, now can also be a single number between 0 and 1. In this case, Bonferroni correction is automatically applied to take into account the multiplicity problem when more than one control chart is applied.
plot_bifd()now allows to choose to produce also contour or perspective plots of
simulate_mfd()is much more general, now it allows to simulate as many covariates as one wants (before the number was fixed to three), it is possible to provide manually the mean and variance function for each variable, it is possible to select the type of correlation function for each variable.
plot_mfd()now relies on patchwork, while the new function
lines_mfd()allows to add new curve to an existing plot.
funchartsnow depends on an older version of R, i.e., >3.6.0 instead of >4.0.0
fof_pc()now is much faster especially when the number of basis functions of the functional coefficient is large since the tensor product has been vectorized.
seedhas been deprecated in all functions, so that reproducibility is achieved by setting externally a seed with
set.seed(), as it is commonly done in R.
sim_funcharts()simulates data sets automatically using the function
simulate_mfd(). The only input required is the sample size for the Phase I, tuning and Phase II data sets.
control_charts_pca()allows automatic selection of components.
get_mfd_array(), with the corresponding real time versions, are now much faster.
seedis deprecated in all functions. Instead, a seed must be set before calling the functions by using
simulate_mfd()simulates example data for
funcharts. It creates a data set with three functional covariates, a functional response generated as a function of the three functional covariates through a function-on-function linear model, and a scalar response generated as a function of the three functional covariates through a scalar-on-function linear model. This function covers the simulation study in Centofanti et al. (2020) for the function-on-function case and also simulates data in a similar way for the scalar response case.
NEWS.mdfile to track changes to the package.
inprod_mfd_diag()calculates the inner product between two multivariate functional data objects observation by observation, avoiding calculating it between all possible couples of observations. Therefore, there are n calculations instead of n2, saving much computational time when calculating the squared prediction error statistic when n is large.
scale_mfd()is pre-computed and therefore is not called many times unnecessarily along the different functions.