展开grid以在R中生成新数据集

Question

我有一个数据框，并希望使用组X和组Y的转换组合从它生成一组新的数据集：

#Group X
df1 <- df+1
df2 <- df-2
df3 <- df*3

#Group Y
df4 <- df*4
df5 <- df^5
df6 <- df/6

我真正想做的是使用expand.grid，然后使用组X和组Y转换的每个组合生成一个新的数据集。(组Y始终在组X之后应用。)这些数据集应该存储在全局环境中。

因此，输出结果将与

df14 <- (df+1)*4
df24 <- (df-2)*4
df34 <- (df*3)*4
df15 <- (df+1)^5
df25 <- (df-2)^5
df35 <- (df*3)^5
df16 <- (df+1)/6
df26 <- (df-2)/6
df36 <- (df*3)/6

我该怎么做呢？(例如，您可以从字面上获取任何数据帧的数据，例如iris。)

如果有帮助，您可以将组X和Y转换重写为函数：

#Group X
Fun1 <- function(x){return(x+1)}
Fun2 <- function(x){return(x-2)}
Fun3 <- function(x){return(x*3)}

#Group Y
Fun4 <- function(x){return(x*4)}
Fun5 <- function(x){return(x^5)}
Fun6 <- function(x){return(x/6)}

我猜对于数据集的名称，像df.Fun1.Fun4这样的名称会更好。(请注意，根据我提供的数据框的名称，df当然应该更改，因此这里将其设置为iris.Fun1.Fun4)

Answer 1

编写一个能做你想做的事情的函数：

foo = function(add, mult) {
  (df + add) * mult
}

然后对add和mult的所需值使用expand.grid，并对其进行迭代。在base中，Map是迭代多个值的一种很好的方式。

params = expand.grid(add = 1:3, mult = 4:6)

df = iris[1:6, 1:3] # numeric sample from iris

result = with(params, Map(foo, add = add, mult = mult))
names(result) = with(params, paste0("add ", add, ", mult ", mult))
result
# $`add 1, mult 4`
#   Sepal.Length Sepal.Width Petal.Length
# 1         24.4        18.0          9.6
# 2         23.6        16.0          9.6
# 3         22.8        16.8          9.2
# 4         22.4        16.4         10.0
# 5         24.0        18.4          9.6
# 6         25.6        19.6         10.8
# 
# $`add 2, mult 4`
#   Sepal.Length Sepal.Width Petal.Length
# 1         28.4        22.0         13.6
# 2         27.6        20.0         13.6
# 3         26.8        20.8         13.2
# 4         26.4        20.4         14.0
# 5         28.0        22.4         13.6
# 6         29.6        23.6         14.8
# 
# $`add 3, mult 4`
#   Sepal.Length Sepal.Width Petal.Length
# 1         32.4        26.0         17.6
# 2         31.6        24.0         17.6
# ...

适配函数而非特定参数：

#Group X
Fun1 <- function(x){return(x+1)}
Fun2 <- function(x){return(x-2)}
Fun3 <- function(x){return(x*3)}

#Group Y
Fun4 <- function(x){return(x*4)}
Fun5 <- function(x){return(x^5)}
Fun6 <- function(x){return(x/6)}

# Put the functions in a list
funs_x = mget(ls(pattern = "Fun[1-3]"))
funs_y = mget(ls(pattern = "Fun[4-6]"))

# iterate over list indices
indices = expand.grid(ind_x = seq_along(funs_x), ind_y = seq_along(funs_y))
result = with(indices, Map(function(ind_x, ind_y) funs_y[[ind_y]](funs_x[[ind_x]](df)), ind_x, ind_y))
names(result) = with(indices, paste("df", names(funs_x)[ind_x], names(funs_y)[ind_y], sep = "."))
result
# $df.Fun1.Fun4
#   Sepal.Length Sepal.Width Petal.Length
# 1         24.4        18.0          9.6
# 2         23.6        16.0          9.6
# 3         22.8        16.8          9.2
# 4         22.4        16.4         10.0
# 5         24.0        18.4          9.6
# 6         25.6        19.6         10.8
# 
# $df.Fun2.Fun4
#   Sepal.Length Sepal.Width Petal.Length
# 1         12.4         6.0         -2.4
# 2         11.6         4.0         -2.4
# 3         10.8         4.8         -2.8
# 4         10.4         4.4         -2.0
# 5         12.0         6.4         -2.4
# 6         13.6         7.6         -1.2
# 
# $df.Fun3.Fun4
#   Sepal.Length Sepal.Width Petal.Length
# 1         61.2        42.0         16.8
# 2         58.8        36.0         16.8
# ...