Cross validation of lambda.ridge — fqpca_cv

Performs cross validation on lambda parameter of fqpca. Only valid if method is set to conquer

Usage

fqpca_cv_lambda(
  data,
  colname = NULL,
  npc = 2,
  pve = NULL,
  quantile.value = 0.5,
  lambda.grid = c(0, 1e-10, 1e-05),
  n.folds = 3,
  return.models = TRUE,
  criteria = "points",
  periodic = TRUE,
  splines.df = 10,
  tol = 0.001,
  max.iters = 20,
  splines.method = "conquer",
  penalized = TRUE,
  verbose.fqpca = FALSE,
  verbose.cv = TRUE,
  seed = NULL
)

Arguments

data: An \((N \times T)\) matrix, a tf object from the tidyfun package or a data.frame containing the functional data as a tf column.
colname: The name of the column containing the functional data. Use only if data is a dataframe and colname is a column in the dataframe.
npc: The number of estimated components.
pve: Float between 0 and 1. Percentage of variability explained by components. This affects the number of components used in the curve reconstruction and error estimation. Set to NULL to avoid this behavior.
quantile.value: The quantile considered.
lambda.grid: Grid of hyper parameter values controlling the penalization on the second derivative of the splines. It has effect only with penalized=TRUE and method='conquer'.
n.folds: Number of folds to be used on cross validation.
return.models: Should the list of all the models built be returned?
criteria: Criteria used to divide the data. Valid values are 'rows', which considers the division based on full rows, or 'points', which considers the division based on points within the matrix.
periodic: Boolean indicating if the data is expected to be periodic (start coincides with end) or not.
splines.df: Degrees of freedom for the splines.
tol: Tolerance on the convergence of the algorithm.
max.iters: Maximum number of iterations.
splines.method: Method used in the resolution of the splines quantile regression model. It currently accepts the methods c('conquer', 'quantreg').
penalized: Boolean indicating if the smoothness should be controlled using a second derivative penalty. This functionality is experimental.
verbose.fqpca: Boolean indicating verbosity of the fqpca function.
verbose.cv: Boolean indicating verbosity of the cross-validation process.
seed: Seed for the random generator number.

Value

A list containing the matrix of scores, the matrix of loadings, a list with all the trained models (if the return_models param is TRUE) and a secondary list with extra information.

Examples


n.obs = 150
n.time = 144

# Generate scores
c1.vals = rnorm(n.obs)
c2.vals = rnorm(n.obs)

# Generate pc's
pc1 = sin(seq(0, 2*pi, length.out = 144))
pc2 = cos(seq(0, 2*pi, length.out = 144))

# Generate data
Y <- c1.vals * matrix(pc1, nrow = n.obs, ncol=n.time, byrow = TRUE) +
c2.vals * matrix(pc2, nrow = n.obs, ncol=n.time, byrow = TRUE)

# Add noise
Y <- Y + matrix(rnorm(n.obs * n.time, 0, 0.4), nrow = n.obs)

# Add missing observations
Y[sample(n.obs*n.time, as.integer(0.2*n.obs*n.time))] <- NA

cv_result <- fqpca_cv_lambda(data=Y, lambda.grid = c(0, 1e-6), n.folds = 2)
#> lambda=0 ---------------------
#> 2026-07-16 19:48:49. Fold: 1
#> 2026-07-16 19:48:49. Fold: 2
#> lambda=0. Execution completed in: 1.081 seconds.
#> lambda=1e-06 ---------------------
#> 2026-07-16 19:48:50. Fold: 1
#> 2026-07-16 19:48:53. Fold: 2
#> lambda=1e-06. Execution completed in: 4.176 seconds.