更改因子水平但保持水平的相同顺序

Question

我有以下数据框：

df <- structure(list(group1 = structure(c(2L, 8L, 5L, 3L, 4L, 6L, 7L, 
                                          9L, 6L, 3L, 9L, 7L, 8L, 4L, 5L, 4L, 9L, 6L, 7L, 8L, 5L, 7L, 8L, 
                                          9L, 5L, 6L, 7L, 6L, 9L, 8L, 9L, 7L, 8L, 9L, 8L, 9L), .Label = c("NEW ENGLAND",
                                          "MIDDLE ATLANTIC", "E. NOR. CENTRAL", "W. NOR. CENTRAL", "SOUTH ATLANTIC",
                                          "E. SOU. CENTRAL", "W. SOU. CENTRAL", "MOUNTAIN", "PACIFIC"), class = "factor"), 
                     Sample_Size_group1 = c(220L, 157L, 372L, 331L, 127L, 135L, 
                                            196L, 267L, 135L, 331L, 267L, 196L, 157L, 127L, 372L, 127L, 
                                            267L, 135L, 196L, 157L, 372L, 196L, 157L, 267L, 372L, 135L, 
                                            196L, 135L, 267L, 157L, 267L, 196L, 157L, 267L, 157L, 267L), 
                     group2 = structure(c(1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 2L, 
                                          2L, 2L, 2L, 2L, 2L, 2L, 3L, 3L, 3L, 3L, 3L, 3L, 4L, 4L, 4L, 
                                          4L, 4L, 5L, 5L, 5L, 5L, 6L, 6L, 6L, 7L, 7L, 8L), .Label = c("NEW ENGLAND", 
                                          "MIDDLE ATLANTIC", "E. NOR. CENTRAL", "W. NOR. CENTRAL", 
                                          "SOUTH ATLANTIC", "E. SOU. CENTRAL", "W. SOU. CENTRAL", "MOUNTAIN", 
                                          "PACIFIC"), class = "factor"),
                     Sample_Size_group2 = c(120L, 120L, 120L, 120L, 120L, 120L, 120L, 120L, 220L, 220L, 220L, 
                                            220L, 220L, 220L, 220L, 331L, 331L, 331L, 331L, 331L, 331L, 
                                            127L, 127L, 127L, 127L, 127L, 372L, 372L, 372L, 372L, 135L, 
                                            135L, 135L, 196L, 196L, 157L)), .Names = c("group1", "Sample_Size_group1", 
                                            "group2", "Sample_Size_group2"), row.names = c(1L, 2L, 3L, 4L, 
                                            5L, 6L, 7L, 8L, 25L, 26L, 27L, 28L, 29L, 30L, 31L, 46L, 47L, 
                                            48L, 49L, 50L, 51L, 64L, 65L, 66L, 67L, 68L, 79L, 80L, 81L, 82L, 
                                            91L, 92L, 93L, 100L, 101L, 106L), class = "data.frame")

group1 和 group2 变量都具有相同顺序的相同因子水平（请注意，并非所有水平都显示在任一变量中）：

levels(df$group1)
[1] "NEW ENGLAND"     "MIDDLE ATLANTIC" "E. NOR. CENTRAL" "W. NOR. CENTRAL" "SOUTH ATLANTIC"  "E. SOU. CENTRAL"
[7] "W. SOU. CENTRAL" "MOUNTAIN"        "PACIFIC" 
levels(df$group2)
[1] "NEW ENGLAND"     "MIDDLE ATLANTIC" "E. NOR. CENTRAL" "W. NOR. CENTRAL" "SOUTH ATLANTIC"  "E. SOU. CENTRAL"
[7] "W. SOU. CENTRAL" "MOUNTAIN"        "PACIFIC"

我想通过将这两个因素与各自的样本量相结合来更改这两个因素的水平，如下所示：

df$newgroup1 <- paste0(df$group1, " (n=", df$Sample_Size_group1, ")")
df$newgroup2 <- paste0(df$group2, " (n=", df$Sample_Size_group2, ")")

这会创建字符向量，因此我想将它们转换为因子：

df$newgroup1 <- factor(df$newgroup1)
df$newgroup2 <- factor(df$newgroup2)

但现在因子水平不再相同，也不是原来的顺序。

levels(df$newgroup1)
[1] "E. NOR. CENTRAL (n=331)" "E. SOU. CENTRAL (n=135)" "MIDDLE ATLANTIC (n=220)" "MOUNTAIN (n=157)"       
[5] "PACIFIC (n=267)"         "SOUTH ATLANTIC (n=372)"  "W. NOR. CENTRAL (n=127)" "W. SOU. CENTRAL (n=196)"
levels(df$newgroup2)
[1] "E. NOR. CENTRAL (n=331)" "E. SOU. CENTRAL (n=135)" "MIDDLE ATLANTIC (n=220)" "MOUNTAIN (n=157)"       
[5] "NEW ENGLAND (n=120)"     "SOUTH ATLANTIC (n=372)"  "W. NOR. CENTRAL (n=127)" "W. SOU. CENTRAL (n=196)"

以下是我希望两个向量的级别：

[1] "NEW ENGLAND (n=120)"     "MIDDLE ATLANTIC (n=220)" "E. NOR. CENTRAL (n=135)" "W. NOR. CENTRAL (n=127)" 
[5] "SOUTH ATLANTIC (n=372)"  "E. SOU. CENTRAL (n=135)" "W. SOU. CENTRAL (n=196)" "MOUNTAIN (n=157)"
[9] "PACIFIC (n=267)"

我知道我可以很容易地手动完成此操作，尤其是使用像 forcats 这样的程序包时，但我如何以编程方式在基础 R 中应用此操作？

Answer 1

这是一个使用 grep 将原始关卡与新的独特元素相匹配的解决方案。之后，按索引对元素进行排序以创建新级别。

# Create newgroup1 and newgroup2
df$newgroup1 <- paste0(df$group1, " (n=", df$Sample_Size_group1, ")")
df$newgroup2 <- paste0(df$group2, " (n=", df$Sample_Size_group2, ")")

# Get the level
LEV <- levels(df$group1)

# Find the unique element in newgroup1 and newgroup2
New_element <- unique(c(df$newgroup1, df$newgroup2))

# Find the index using grep
index <- sapply(LEV, grep, x = New_element)

# Use New_element[index] to set new level
df$newgroup1 <- factor(df$newgroup1, levels = New_element[index])
df$newgroup2 <- factor(df$newgroup2, levels = New_element[index])

# Check the results
levels(df$newgroup1)
levels(df$newgroup2)

Answer 2

使用 unique 将级别彼此相邻排列，按原始因子排序，然后使用该顺序定义新因子：

group1_levels = unique(df[, c("group1", "newgroup1")])
group1_levels = group1_levels[order(group1_levels$group1), ]

df$newgroup1 = factor(df$newgroup1, levels = group1_levels$newgroup1)

group2_levels = unique(df[, c("group2", "newgroup2")])
group2_levels = group2_levels[order(group2_levels$group2), ]

df$newgroup2 = factor(df$newgroup2, levels = group2_levels$newgroup2)

更改因子水平但保持水平的相同顺序

Making changes to factor levels but keeping the same order of the levels

r

r-factor