Skip to contents

Hybrid Data Constructor (hdClass)

hdClass.Rd

Constructs an object of class hdClass, representing hybrid data composed of multiple functional and/or raw variables.

Usage

hdClass(hdlist, argval = NULL, Smoothing_parameter = 0, Sparsity_parameter = 0)

Arguments

hdlist

A list of fdClass, rdClass, or imgClass objects.

argval

Optional numeric vector of grid points along the rows. If NULL, it defaults to a uniform grid over 0 and 1.

Smoothing_parameter

Smoothing parameter(s) for the rows:

  • If 0, no smoothing is applied.

  • If a numeric value or vector, it is used directly.

  • If NULL, defaults to 2^seq(-30, 5, length.out = 10) for tuning.

Sparsity_parameter

Sparsity parameter(s) for the rows:

  • If 0, no sparsity is applied.

  • If a numeric vector, it is used for tuning.

  • If NULL, a suitable default sequence is generated.

Value

An object of class hdClass (a matrix) with several hybrid-aware attributes.

Details

The hdClass is an S3 class for representing hybrid data, consisting of a list of functional (fdClass), raw (rdClass), or image (imgClass) data objects.

  • The input list (hdlist) can contain a mix of fdClass, rdClass, and imgClass objects.

  • All objects must have the same number of observations (i.e., same number of rows).

  • Image objects (imgClass) are interpreted as either functional or raw based on their internal class inheritance.

The resulting object is a matrix with column-wise concatenation of the input matrices and includes the following attributes:

  • GridPoints_u: Row grid points.

  • Smoothing_parameter: Row smoothing parameter(s).

  • Sparsity_parameter: Row sparsity parameter(s).

  • n_var: Number of variables (i.e., number of objects in hdlist).

  • ncol: Number of columns in each variable.

  • Smoothing_parameter_col: List of smoothing parameters for each variable.

  • Sparsity_parameter_col: List of sparsity parameters for each variable.

  • GridPoints_v: List of grid points (columns) for each variable.

  • variable_types: Character vector of type labels for each variable: "hd" (functional) or "rd" (raw).

Examples

fd_obj <- fdClass(matrix(rnorm(100), nrow = 10))
rd_obj <- rdClass(matrix(rnorm(100), nrow = 10))

hd_obj <- hdClass(
  hdlist = list(fd_obj, rd_obj),
  argval = seq(0, 1, length.out = 10),
  Smoothing_parameter = 0.5,
  Sparsity_parameter = 2
)

print(hd_obj)
#> Hybrid Data (hdClass) Object
#> ===================================
#> Dimensions           : 10 rows x 20 columns
#> Number of Variables  :  2 
#> Columns per Variable :  10 | 10 
#> -----------------------------------
#> Grid Points (Rows)   : 0, 0.111111111111111, 0.222222222222222 , ...,  0.888888888888889, 1 
#> Grid Points (Columns):
#>   - Variable 1: 0.1, 0.2, 0.3 , ...,  0.9, 1 
#>   - Variable 2: NULL
#> Smoothing Parameter (Row): 0.5 
#> Smoothing Parameters (Col):
#>   - Variable 1: 0 
#>   - Variable 2: 0 
#> Sparsity Parameter (Row): 2 
#> Sparsity Parameters (Col):
#>   - Variable 1: 0 
#>   - Variable 2: 0 
#> ===================================
#> First few rows and columns of the data:
#>             V1          V2         V3         V4          V5
#> [1,] 0.1065870 -0.31031012  0.3038483  0.1216206 -0.95362873
#> [2,] 1.1769142 -0.02010818 -0.1833885 -1.0780673  0.41626080
#> [3,] 0.4473912 -2.39020034  0.5596497 -1.1435657  0.11402961
#> [4,] 2.2729548  0.88986536 -0.1865538 -0.5296437  0.06391875
#> [5,] 0.1360582 -1.48281332 -0.8122754 -0.6812732 -0.91933224
attr(hd_obj, "variable_types")  # Shows "hd", "rd", etc.
#> [1] "hd" "rd"