Fold multiple columns into a single list-column

Combines multiple related columns (e.g., Fourier coefficients A1, A2, ..., B1, B2, ...) into a single list-column where each row contains a numeric vector of all the values. This is useful for storing morphometric coefficients compactly in a tibble.

Usage

fold(.data, ..., .class = NULL)

Arguments

.data

A data frame or tibble

...

A single named argument using tidyselect syntax. The name becomes the new list-column name, and the selection specifies which columns to fold. Must be exactly one named argument.

.class

Character vector of classes to add to the list-column and individual vectors. If NULL (default), no classes are added - the list-column remains a plain list. For morphometric coefficients, specify classes explicitly, e.g., .class = c("eft", "coe") or .class = "coe".

Value

A tibble with selected columns removed and replaced by a single list-column containing numeric vectors (one per row). If .class is provided, both the list-column and individual vectors get those classes (e.g., c("eft", "coe", "list") and c("eft", "coe", "numeric")). If .class = NULL, returns a plain list-column.

Details

The function:

1. Selects columns using tidyselect (e.g., A1:D5, starts_with("harm"))

2. Converts them to a matrix

3. Splits the matrix by rows to create a list of numeric vectors

4. Preserves original column names as names in each vector

5. If .class is provided, adds classes to both the list-column and individual vectors

6. Removes the original columns and adds the new list-column

This is particularly useful after Fourier analysis where you have many harmonic coefficient columns that you want to store together. For morphometric coefficients, always specify .class explicitly to ensure proper class attributes.

See also

unfold() for the reverse operation, get_coe_cols() for detection

Examples

# Create sample data with coefficient columns
df <- tibble::tibble(
  id = 1:3,
  A1 = c(1, 2, 3),
  A2 = c(4, 5, 6),
  B1 = c(7, 8, 9),
  B2 = c(10, 11, 12)
)

# Fold coefficient columns into a single list-column
df_folded <- fold(df, coe = A1:B2)  # Plain list, no classes
df_folded$coe[[1]]
#> A1 A2 B1 B2 
#>  1  4  7 10 
#    A1    A2    B1    B2
#     1     4     7    10

# For morphometric coefficients, specify classes explicitly
df_eft <- fold(df, coe = A1:B2, .class = c("eft", "coe"))

# List-column has classes only if specified
class(df_eft$coe)
#> [1] "eft"  "coe"  "list"
# [1] "eft"  "coe"  "list"

# Each element also has classes
class(df_eft$coe[[1]])
#> [1] "eft"     "coe"     "numeric"
# [1] "eft"     "coe"     "numeric"
names(df_eft$coe[[1]])
#> [1] "A1" "A2" "B1" "B2"
# [1] "A1" "A2" "B1" "B2"

# Without .class, just a plain list
df_plain <- fold(df, scores = A1:B2)
class(df_plain$scores)
#> [1] "array"
# [1] "list"