首页 文章

如何根据数据的整体顺序更改特定的分类变量

提问于
浏览 次
3

我每五天收集一次关于植物发育或物候学的数据（使用分类变量“代码”编码），沿着横断面划分为78个连续区段 . 每个物种都在每个区段的横断面上进行调查 .

我的研究重复了100年前的历史研究，我保留了最初的物候编码方案，但没有考虑如何在夏天之后分析数据！

我在收集数据时没有考虑的问题是代码遵循一个序列，其中一个代码在夏天的早晚出现 . 具体来说，代码是：

b1 = single flower
b2 = sparse flowers (two or three)
b3 = flowers common (more than three)
B4 = flowering ended

根据原始研究的方法，在夏季为任何开花植物收集的代码序列将类似于b1，b2，b3，b2，b1，b4 . 请注意，我们每隔五天访问样带，并且代码可能在连续几天内重复，例如b1，b1，b2，b2，b2，b2，b3，b3，b3，b2，b2，b1，b4 .

我想重新编码'b1'和'b2'代码如下（参见示例和示例数据）：

1.如果'b1'出现在'b2'或'b3'之前那么它应该是'b1a'并且如果它出现在'b2'或'b3'之后那么它应该是'b1b' . 请注意，有时在观察序列中没有'b2'或'b3' .

2.如果'b2'发生在'b3'之前那么它应该是'b2a'，如果它发生在'b3'之后它应该是'b2b' . 或者如果没有'b3'那么'b2'应该是'b2a' . 请注意，重要的是要记住，在最后一次出现'b3'之后，可能会有多次'b2'的观察（参见示例和示例数据） .

3.考虑'b1'和'b2'可能在没有和'b3'的情况下发生，在这种情况下，两者都会被编码为'b1a'和'b2a' .

以下是数据的样子：

Date    Segment Species Code
01-Jun-17   1   A   b1
06-Jun-17   1   A   b1
10-Jun-17   1   A   b2
14-Jun-17   1   A   b2
19-Jun-17   1   A   b2
23-Jun-17   1   A   b3
28-Jun-17   1   A   b3
03-Jul-17   1   A   b2
08-Jul-17   1   A   b2
14-Jul-17   1   A   b1
19-Jul-17   1   A   b4
23-Jul-17   1   A   b4

它应该是这样的：

Date    Segment Species Code
01-Jun-17   1   A   b1
06-Jun-17   1   A   b1a
10-Jun-17   1   A   b2a
14-Jun-17   1   A   b2a
19-Jun-17   1   A   b2a
23-Jun-17   1   A   b3
28-Jun-17   1   A   b3
03-Jul-17   1   A   b2b
08-Jul-17   1   A   b2b
14-Jul-17   1   A   b1b
19-Jul-17   1   A   b4
23-Jul-17   1   A   b4

以下是示例数据：

Test.Data<- structure(list(Date = structure(c(17318, 17323, 17327, 17331, 
17336, 17340, 17345, 17350, 17355, 17361, 17366, 17318, 17323, 
17327, 17331, 17336, 17340, 17345, 17350, 17355, 17361, 17366, 
17370, 17375, 17318, 17323, 17327, 17331, 17336, 17340, 17345, 
17350, 17355, 17361, 17366, 17318, 17323, 17327, 17331, 17336, 
17340, 17345, 17350, 17355, 17361, 17366, 17370, 17375, 17355, 
17361, 17366, 17370, 17375, 17350, 17355, 17361, 17366, 17370
), class = "Date"), Segment = c(1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2, 2, 
2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 
2, 1, 1, 1, 1, 1), Species = c("A", "A", "A", "A", "A", "A", 
"A", "A", "A", "A", "A", "B", "B", "B", "B", "B", "B", "B", "B", 
"B", "B", "B", "B", "B", "B", "B", "B", "B", "B", "B", "B", "B", 
"B", "B", "B", "A", "A", "A", "A", "A", "A", "A", "A", "A", "A", 
"A", "A", "A", "C", "C", "C", "C", "C", "C", "C", "C", "C", "C"
), Code = c("b1", "b1", "b2", "b2", "b2", "b3", "b3", "b2", "b2", 
"b4", "b4", "b1", "b2", "b2", "b2", "b3", "b3", "b3", "b2", "b2", 
"b2", "b1", "b4", "b4", "b1", "b1", "b2", "b2", "b2", "b3", "b3", 
"b2", "b2", "b4", "b4", "b1", "b2", "b2", "b2", "b3", "b3", "b3", 
"b2", "b2", "b2", "b4", "b4", "b4", "b3", "b3", "b2", "b1", "b4", 
"b1", "b1", "b2", "b2", "b4")), .Names = c("Date", "Segment", 
"Species", "Code"), row.names = c(NA, -58L), class = "data.frame")
r dataframe data.table

2 回答

  • 4

    使用data.table：

    library(data.table)
setDT(Test.Data)
Test.Data[, temp := rleid(Code), by = .(Segment, Species)] #unique ids for the sequence of codes
Test.Data[Code == "b2", Code := paste0(Code, letters[rleid(temp)]), 
  by = .(Segment, Species)] #use the unique ids inside subset
Test.Data[, temp := NULL]
#          Date Segment Species Code
# 1: 2017-06-01       1       A   b1
# 2: 2017-06-06       1       A   b1
# 3: 2017-06-10       1       A  b2a
# 4: 2017-06-14       1       A  b2a
# 5: 2017-06-19       1       A  b2a
# 6: 2017-06-23       1       A   b3
# 7: 2017-06-28       1       A   b3
# 8: 2017-07-03       1       A  b2b
# 9: 2017-07-08       1       A  b2b
#10: 2017-07-14       1       A   b4
#11: 2017-07-19       1       A   b4
#12: 2017-06-01       1       B   b1
#13: 2017-06-06       1       B  b2a
#14: 2017-06-10       1       B  b2a
#15: 2017-06-14       1       B  b2a
#16: 2017-06-19       1       B   b3
#17: 2017-06-23       1       B   b3
#18: 2017-06-28       1       B   b3
#19: 2017-07-03       1       B  b2b
#20: 2017-07-08       1       B  b2b
#21: 2017-07-14       1       B  b2b
#</cont>
    回复于
  • 2

    你可以使用 dplyr 包

    library(dplyr)
Test.Data %>% 
  group_by(Species) %>% 
  mutate(hadb3 = cumsum(Code=="b3")>0) %>%
  mutate(Code = ifelse(Code=="b2" & !hadb3,"b2a",Code)) %>% 
  mutate(Code = ifelse(Code=="b2" & hadb3,"b2b",Code))

    结果：

    # A tibble: 48 x 5
# Groups:   Species [2]
         Date Segment Species  Code hadb3
       <date>   <dbl>   <chr> <chr> <lgl>
 1 2017-06-01       1       A    b1 FALSE
 2 2017-06-06       1       A    b1 FALSE
 3 2017-06-10       1       A   b2a FALSE
 4 2017-06-14       1       A   b2a FALSE
 5 2017-06-19       1       A   b2a FALSE
 6 2017-06-23       1       A    b3  TRUE
 7 2017-06-28       1       A    b3  TRUE
 8 2017-07-03       1       A   b2b  TRUE
 9 2017-07-08       1       A   b2b  TRUE
10 2017-07-14       1       A    b4  TRUE
# ... with 38 more rows

    mutate(hadb3 = cumsum(Code=="b3")>0) 创建一个逻辑列，用于检查 b3 之前是否已出现，并且足以使用ifelse语句获取结果 .

    回复于

相关问题