在 R 中使用 expand.grid 在 y 集中创建所有可能的 x 因子组合

Question

是否可以在 R 中使用 expand.grid() 在 y 集合中创建 x 因子的所有可能组合？

例如，我有 12 个因素：

Factor1 = c("1", "2", "3", "4"),       #Fixed Attribute: 4 lvls
Factor2 = c("5", "6", "7", "8", "9"),  #Fixed Attribute: 5 lvls
Factor3 = c("10", "11", "12","13"),    #Fixed Attribute: 4 lvls
Factor4 = c("14", "15", "16"),         #Fixed Attribute: 4 lvls
Factor5 = c("17", "18", "19", "20", "21"),  #Variable Attribute: 5 lvls
Factor6 = c("22", "23"),                    #Variable Attribute: 2 lvls
Factor7 = c("24", "25", "26"),              #Variable Attribute: 3 lvls
Factor8 = c("27", "28", "29")               #Variable Attribute: 3 lvls
Factor9 = c("30", "31", "32", "33"),        #Variable Attribute: 4 lvls
Factor10= c("34", "35"),                    #Variable Attribute: 2 lvls
Factor11 = c("36", "37", "38"),             #Variable Attribute: 3 lvls
Factor12 = c("39", "40", "41")              #Variable Attribute: 3 lvls

我希望始终在expand.grid() 中包含前 4 个（即它们是固定的），并在所有可能的 4 个集合中循环到最后 8 个，这等于 70 个唯一集合。然后附加所有生成的 70 个数据帧。

我可以通过创建 70 个不同的 expand.grid() 代码块来以蛮力的方式实现这一点，但是有没有一种技术上不太优雅的方式来做到这一点？

例如暴力破解方式如下：

expand.grid(Factor1, Factor2,Factor3,Factor4,Factor5,Factor6,Factor7,Factor8)
expand.grid(Factor1, Factor2,Factor3,Factor4,Factor5,Factor6,Factor7,Factor9)
expand.grid(Factor1, Factor2,Factor3,Factor4,Factor5,Factor6,Factor7,Factor10)
expand.grid(Factor1, Factor2,Factor3,Factor4,Factor5,Factor6,Factor7,Factor11)
expand.grid(Factor1, Factor2,Factor3,Factor4,Factor5,Factor6,Factor7,Factor12)
....etc...

所以我最终会得到 70 个不同的数据框，因为有 70 种独特的方法可以从因子 4-12 中选择 4 个因子（即，有 70 种方法可以从 8 个列表中选择 4 个项目）

此外，生成的数据框可能是 150 万行。这会导致内存问题吗？

谢谢，

Answer 1

如果我理解你的话，这应该做你想要的：

l <- list(
    Factor1 = c("1", "2", "3", "4"),       #Fixed Attribute: 4 lvls
    Factor2 = c("5", "6", "7", "8", "9"),  #Fixed Attribute: 5 lvls
    Factor3 = c("10", "11", "12","13"),    #Fixed Attribute: 4 lvls
    Factor4 = c("14", "15", "16"),         #Fixed Attribute: 4 lvls
    Factor5 = c("17", "18", "19", "20", "21"),  #Variable Attribute: 5 lvls
    Factor6 = c("22", "23"),                    #Variable Attribute: 2 lvls
    Factor7 = c("24", "25", "26"),              #Variable Attribute: 3 lvls
    Factor8 = c("27", "28", "29"),               #Variable Attribute: 3 lvls,
    Factor9 = c("30", "31", "32", "33"),        #Variable Attribute: 4 lvls
    Factor10= c("34", "35"),                    #Variable Attribute: 2 lvls
    Factor11 = c("36", "37", "38"),             #Variable Attribute: 3 lvls
    Factor12 = c("39", "40", "41")              #Variable Attribute: 3 lvls
)



# Get the names of the other 8
others <- names(l)[-c(1:4)]
# Get names of the 4 fixed ones
fixed <- names(l)[1:4]

# Get all combinations of 4 of names of the others
combos <- combn(others, 4)

# Get the list of 70 expand grid outputs of combinations (fixed, combo_of_4)
out <- apply(combos, 2, function(x) expand.grid(l[c(fixed,x)]))