如何用X、Y向量创建带条件过滤的公式列表/tibble供map函数使用

Absolutely! This is totally doable with rep(), conditional filtering, reformulate(), and purrr::map() to tie it all together. Let's walk through a step-by-step solution that gives you either a list of tibbles (each length 9) or a formula list ready for mapping.

First, let's start with your base vectors:

X <- c("A", "B", "C")
Y <- c("X", "L", "Z")

Step 1: Build a conditional filtering & processing function #

We'll create a function that takes each value from Y, filters X based on your rules, generates the repeated 9-length vector, and creates the formula with reformulate():

library(purrr)
library(tibble)
library(dplyr)

process_y_element <- function(y_val) {
  # Filter X based on Y value
  filtered_x <- case_when(
    y_val == "L" ~ X[X != "C"],  # Remove "C" when Y is "L"
    y_val == "Z" ~ X[X != "B"],  # Remove "B" when Y is "Z"
    TRUE ~ X                     # Keep full X for Y = "X"
  )
  
  # Generate 9-length repeated X (uses length.out to guarantee exact length)
  repeated_x <- rep(filtered_x, length.out = 9)
  
  # Create the formula: Y value as response, filtered X terms as predictors
  # We use unique(filtered_x) because formulas don't need repeated variable names
  formula <- reformulate(termlabels = unique(filtered_x), response = y_val)
  
  # Package everything into a tibble (easy to work with)
  tibble(
    y = rep(y_val, 9),
    x = repeated_x,
    formula = list(formula)  # Store formula as a list column to avoid flattening
  )
}

Step 2: Generate your desired output #

Now use map() to apply this function to every element in Y:

Option 1: List of tibbles (3 elements, each length 9)

result_list <- map(Y, process_y_element)

# Check the first element (Y = "X")
result_list[[1]]
#> # A tibble: 9 × 3
#>   y     x     formula    
#>   <chr> <chr> <list>     
#> 1 X     A     <formula>  
#> 2 X     B     <formula>  
#> 3 X     C     <formula>  
#> 4 X     A     <formula>  
#> 5 X     B     <formula>  
#> 6 X     C     <formula>  
#> 7 X     A     <formula>  
#> 8 X     B     <formula>  
#> 9 X     C     <formula>

Option 2: Single combined tibble (length 27)

If you prefer one unified table, use map_dfr():

result_tibble <- map_dfr(Y, process_y_element)

Option 3: Just the formula list (ready for map operations)

If you only need the formulas without the tibble wrapper, simplify the code:

formula_list <- map(Y, function(y_val) {
  filtered_x <- case_when(
    y_val == "L" ~ X[X != "C"],
    y_val == "Z" ~ X[X != "B"],
    TRUE ~ X
  )
  reformulate(unique(filtered_x), y_val)
})

# View the formulas
formula_list
#> [[1]]
#> X ~ A + B + C
#> <environment: 0x0000022a1e8f7d58>
#> 
#> [[2]]
#> L ~ A + B
#> <environment: 0x0000022a1e8f7d58>
#> 
#> [[3]]
#> Z ~ A + C
#> <environment: 0x0000022a1e8f7d58>

Quick notes: #

• Using rep(filtered_x, length.out = 9) is cleaner than calculating n manually—it automatically repeats the filtered vector enough times and truncates/pads to hit exactly 9 elements.
• If you really need repeated variable names in your formulas (uncommon for statistical modeling, but possible for other use cases), just swap unique(filtered_x) with repeated_x in the reformulate() call.

内容的提问来源于stack exchange，提问作者Geet