ChangeLog

v 6.6

1. Update the ftsmPI function
2. Update the nipals.R function

v 6.5

1. Move the One_way_median_polish and Two_way_median_polish to a newly created hdftsa package

v 6.4

1. Add One_way_median_polish and One_way_Residuals functions
2. Remove the vignette

v 6.3

1. Add Two_way_Residuals_means and Two_way_Residuals functions
2. Update the forecast.ftsm function

v 6.2

1. Add sieve bootstrapping
2. Update the mftsc.R code
3. Add FANOVA and two-way functional median polish decompositions
4. Add the all_hmd_female_data and all_hmd_male_data data sets

v 6.1

1. Add CoDa_FPCA, LQDT_FPCA, Horta_Ziegelmann_FPCA and skew_t_fun functions for forecasting densities
2. Add CoDa_BayesNW function for forecasting densities
3. Add GAEVforecast function for forecasting a time series of densities of extreme values

v 6.0

1. Update the T_stationary function by introducing critical values (see Table 6.1 in Horvath and Kokoszka's book)

v 5.9

1. Add the plsPI_var.R function
2. Update the dynamic_FLR.R function
3. Update the dmfpca.R function
4. Update the pm_10_GR.Rd manual file

v 5.8

1. Add MAF_Multivariate function to perform maximum autocorrelation factors
2. Add multilevel functional principal component analysis and multiple functional time series clustering
3. Update long_run_covariance_estimation function

v 5.7

1. Add a function (ER_GR) for selecting the number of retained principal components
2. Add a function (stop_time_detect) for detecting the optimal stopping time in a curve time series
3. Add a function (stop_time_sim_data) for generating simulated data following a functional autoregression of order one with a breakpoint at the middle of a curve sample

v 5.6

1. Update the fdpca and rapca functions in R
2. Update the farforecast manual and farforecast function in R

v 5.5

1. Update fdpca and ftsm functions
2. Add ftsa:::method.FPE function used in the farforecast function
3. Add dmfpca function for implementing dynamic multilevel functional principal component analysis
4. Add mftsc function for clustering multiple functional time series

v 5.4

1. Update the ElNino data set

v 5.3

1. Add the high-dimensional functional principal component regression
2. Add the forecast function for the high-dimensional functional principal component regression
3. Add the facf function for computing the functional autocorrelation function at various lags
4. Update the two vignettes' titles

v 5.2

1. Update the ftsm function to fix the warning message from the fts function
2. Update the forecast.ftsm function to fix the warning message from the fts function
3. Add some new references to the ../man

v 5.1

1. Replace Depends: R (>= 3.5.0) to R (>= 3.4.0)

v 5.0

1. Remove R/diff.R
2. Remove R/diff.default.R

v 4.9

1. Update the forecast.ftsm.R (model type ets.na in forecast.ftsm now returns en enriched arima model)
2. Update the forecast.ftsm2.R (replaced rw and rwdrift models internally with the corresponding arima model)
3. Update the ftsm.R (ftsm can handle years with only NAs)
4. Update the pegelsna.R (ets.na accepts a lower bound for the parameters)
5. Update the plot.ftsf.R (plot the fitted lines as well)
6. Update the plot.ftsm.R (make the ftsm plots splittable by introducing a new parameter component)
7. Update the ftsm.Rd (fixed typo in the help page)

v 4.8

1. Add long_run_covariance_estimation function
2. Add kweights and FlatTop functions
3. Update smdape.R and smape.R functions
4. remove median re-definition and create an S3 method for median.fts
5. remove sd and var re-definition and create their S3 methods for var.fts and sd.fts

v 4.7

1. Add a new vignette to describe the use of farforecast and T_stationary functions

v 4.6

1. Update the farforecast function and its help file
2. Update the T_stationary function and its help file

v 4.5

1. Update the description of the ftsa-package
2. Update the T_stationary function
3. Update the farforecast function
4. Update the ftsa:::ftsmPI function for forecast horizon h > 1
5. Add the meboot option in the pcscorebootstrapdata for bootstrapping a functional time series
6. Add the dynamic_FLR function to update point and interval forecasts via a functional linear regression

v 4.4

1. Include PM10 pollution data set and its square root transformation
2. Include multivariate time series method for forecasting principal component scores, called farforecast
3. Include a stationarity test, called T_stationary, for functional time series
4. Update forecast.ftsm function (ftsa:::ftsmPI)

v 4.3

1. Remove R2jags package dependency and put it as a suggested package

v 4.2

1. Update dynupdate function
2. Update ftsm function
3. Update fdpca function
4. Update plsPI function
5. Add MFDM function

v 4.1

1. Add pcaPP and MASS packages as imports
2. Remove \donttest in the examples
3. Add \alias{ftsa} in the ftsa-package help file

v 4.0

1. Update the ftsm and forecast.ftsm functions for handling functional time series

v 3.9

1. Update the plot.ftsf function
2. Add Imports field in the description file

v 3.8

1. Update the ftsa:::fdpca function
2. Update the ftsm function
3. Update the plot.ftsm function

v 3.7

1. Update the forecast.ftsm function
2. Update the forecast.ftsm2 function

v 3.6

1. Add more detail to the example section
2. Add the depth.radius option to the mean.fts, median.fts, sd.fts and var.fts

v 3.5

1. Update the plot.fmres function using the colorspace package
2. Delete some internal functions, namely cm.spline.R, KalmanLike.R, L1median.R, PCAproj.R, ash2.R, bin2.R

v 3.4

1. Update the ftsm function
2. Update the var.default function
3. Update the ftsa:::plsPI function
4. Add the vignette

v 3.3

1. Update the sd.default.R, var.default.R functions.

v 3.2

1. Delete the standard sd.R, var.R, quantile.R, median.R functions, which are available in base

v 3.1

1. Update the scaled error measures in the error function

v 3.0

1. Correct the plot.ftsm function
2. Update plot.ftsf function

v 2.9

1. Correct the plot.ftsm function

v 2.8

1. Update the plot.ftsm function
2. Add the forecast.fplsr function
3. Add the forecast.ftsmiterative function
4. Add the plot.fplsr function
5. Add the pcscorebootstrapdata function

v 2.7

1. Update the extract function
2. Add the ftsmweightselect function to select the optimal weighted used in the weighted ftsm

v 2.6

1. Update ftsm function to produce the orthonormal basis functions

v 2.5

1. correct ftsm function

v 2.4

1. correct ftsm function weight

v 2.3

1. correct ftsm function

v 2.2

1. correct ftsm function

v 2.1

1. update ftsm
2. update mean.fts and median.fts

v 2.0

1. update plot.ftsm
2. add varprop output in ftsm function when weight = TRUE

v 1.9

1. Remove some duplicate functions

v 1.8

1. Update the arfima function in forecast.ftsm

v 1.7

1. Change the package dependency of fds
2. Update the forecast.ftsm() function (allowing struct and arfima models, and fixing the way stationary models are used)
3. Update the isfe.fts() function (adding ar and arfima as possible forecast methods)
4. Make plot.ftsm visible
5. Update plot.ftsm to allow plotting specific components

v 1.6

1. Update the JCGS reference

v 1.5

1. Update the ftsa:::me function
2. Add the ftsa:::mpe function

v 1.4

1. Update dynupdate function to incorporate three ways of computing functional principal component analysis.

v 1.3

1. Add dynupdatelambdaselect function.

v 1.2

1. Correct the extract function
2. Add ftsmweightselect and fplsrweightselect functions to determine geometrically decaying weights

v 1.1

1. Update the references
2. Correct the yname of a functional object