將因子級別更改為R中具有不同順序的數字

Question

我的數據集中有一個與以下因素相似的列：

response = c("Met Expectations", "Exceeded Expectations", "Exceeded    Expectations", "Unacceptable", NA, "Did not meet Expectations" )
factor(response, levels = c("Exceeded Expectations", "Met Expectations", "Did not meet Expectations", "Unacceptable"))

這就是我得到的關卡。 我已經按照在數據集中獲取的方式進行了排序。

[1] Met Expectations          Exceeded Expectations     Exceeded Expectations     Unacceptable             
[5] <NA>                      Did not meet Expectations
Levels: Exceeded Expectations Met Expectations Did not meet Expectations Unacceptable

我想將這些級別轉換為數字。 所以我嘗試了這個：

as.numeric(factor(response, levels = c("Exceeded Expectations", "Met Expectations", "Did not meet Expectations", "Unacceptable")))

我得到正確的輸出：

[1]  2  1  1  4 NA  3

但是我希望順序是4 3 2 1而不是1 2 34。我期望：

Exceeding Expectation to be 4
Met Expectations to be 3
Did not meet Expectations to be 2
Unacceptable to be 1
[1] 3  4  4  1  NA  2

如何更改訂單？ 有沒有直接的方法可以做到這一點，否則我可以在將數字轉換為數字后修改數字。 在此示例中，我可以更改級別參數的順序並獲得所需的內容，但是由於我的數據集按照以下特定順序為我提供了級別：

級別：超出期望達到期望未達到期望不可接受

我正在嘗試找出一種簡單的方法。

Answer 1

當您創建因子時，請更改為levels = rev(c("Exceeded Expectations", "Met Expectations", "Did not meet Expectations", "Unacceptable")) 。 rev將反轉級別的順序，從而反轉分配給類別標簽的數值的順序。

response = c("Met Expectations", "Exceeded Expectations", "Exceeded Expectations", "Unacceptable", NA, "Did not meet Expectations")

response = factor(response, levels = c("Exceeded Expectations", "Met Expectations", "Did not meet Expectations", "Unacceptable"))

as.numeric(response)

#[1]  2  1  1  4 NA  3

response = factor(response, levels = rev(c("Exceeded Expectations", "Met Expectations", "Did not meet Expectations", "Unacceptable")))

as.numeric(response)

#[1]  3  4  4  1 NA  2

如果您不想更改因子變量中水平的順序，則可以執行以下操作：

response = factor(response, levels = c("Exceeded Expectations", "Met Expectations", "Did not meet Expectations", "Unacceptable"))

# Convert to numeric with numeric order of factor reversed
5 - as.numeric(response)

#[1]  3  4  4  1 NA  2

如果級別數可以變化，您還可以動態計算級別數：

# Convert to numeric with numeric order of factor reversed
length(unique(na.omit(response))) + 1 - as.numeric(response)