如何从R中带有描述的变量中提取月份和年份？

Question

我有一个包含不同类型变量的数据集，其中 1 个变量包含带有年份和月份的描述，我想从该变量中提取月份和年份，但我无法获取。

Sample_Data

var1   var2 
203    UPFRONT BROKERAGE FOR THE MONTH OF MARCH 2008                           
205    UPFRONT BROKERAGE FOR THE MONTH OF MAY 2010                           
206    2008 MARCH MONTH BROKERAGE                            
207    UPFRONT BROKERAGE FOR 2009 MONTH OF APRIL                           
204    BROKERAGE FOR THE MONTH OF MARCH 2008                           


Expected_output:

var1   var2                                            month   year     
203    UPFRONT BROKERAGE FOR THE MONTH OF MARCH 2008   MARCH   2008                      
205    UPFRONT BROKERAGE FOR THE MONTH OF MAY 2010     MAY     2010                           
206    2008 MARCH MONTH BROKERAGE                      MARCH   2008                      
207    UPFRONT BROKERAGE FOR 2009 MONTH OF APRIL       APRIL   2009                           
204    BROKERAGE FOR THE MONTH OF MARCH 2008           MARCH   2008

Tried:
library(lubridate)
Sample_Data$month = month(Sample_Data$var2)
Sample_Data$year = year(Sample_Data$var2)

我尝试了不同的方法，例如使用 lubridate、posixlt 但无法找到解决方案。 请以这种方式帮助我。

Answer 1

我们可以通过指定正则表达式来匹配输入数据集中显示的字符，从而使用从tidyr extract 。

library(tidyr)
extract(df1, var2, into=c('month', 'year'), '.*\\s+([A-Z]+)\\s+(\\d+)$', 
             remove=FALSE, convert=TRUE)
#  var1                                          var2 month year
#1  203 UPFRONT BROKERAGE FOR THE MONTH OF MARCH 2008 MARCH 2008
#2  205   UPFRONT BROKERAGE FOR THE MONTH OF MAY 2010   MAY 2010
#3  206 UPFRONT BROKERAGE FOR THE MONTH OF MARCH 2008 MARCH 2008
#4  207 UPFRONT BROKERAGE FOR THE MONTH OF APRIL 2009 APRIL 2009
#5  204 UPFRONT BROKERAGE FOR THE MONTH OF MARCH 2008 MARCH 2008

---

或者使用base R ，我们从 'var2' 中的字符串开头删除子字符串，捕获单词 ( \\\\w+ ) 后跟空格 ( \\\\s+ ) 后跟数字 ( \\\\d+ ) 直到结束字符串，在替换中，我们指定捕获组（ \\\\1 ）。 我们使用read.table读取此内容以在 'df1' 中创建新列。

df1[c('month', 'year')] <-  read.table(text=sub('.*(\\b\\w+\\s+\\d+)$',
                                   '\\1', df1$var2), stringsAsFactors=FALSE)
df1
#  var1                                          var2 month year
#1  203 UPFRONT BROKERAGE FOR THE MONTH OF MARCH 2008 MARCH 2008
#2  205   UPFRONT BROKERAGE FOR THE MONTH OF MAY 2010   MAY 2010
#3  206 UPFRONT BROKERAGE FOR THE MONTH OF MARCH 2008 MARCH 2008
#4  207 UPFRONT BROKERAGE FOR THE MONTH OF APRIL 2009 APRIL 2009
#5  204 UPFRONT BROKERAGE FOR THE MONTH OF MARCH 2008 MARCH 2008

注意：在这两种方法中，我们都将新列转换为它们各自的class 。

数据

df1 <- structure(list(var1 = c(203L, 205L, 206L, 207L, 204L),
var2 = c("UPFRONT BROKERAGE FOR THE MONTH OF MARCH 2008", 
"UPFRONT BROKERAGE FOR THE MONTH OF MAY 2010",
"UPFRONT BROKERAGE FOR THE MONTH OF MARCH 2008", 
"UPFRONT BROKERAGE FOR THE MONTH OF APRIL 2009",
"UPFRONT BROKERAGE FOR THE MONTH OF MARCH 2008"
)), .Names = c("var1", "var2"), class = "data.frame", 
row.names = c(NA, -5L))

Answer 2

您还不能完全将其视为日期，因为您需要解析字符串。 尝试t(sapply(strsplit(Sample_Data$var2," "),function(x) x[7:8]))得到你想要的两列。

Answer 3

您不需要 lubridate 因为您并没有真正使用 Date 数据类型。 在 base 中使用strsplit将您的var2拆分为“单词”。 看起来月份总是倒数第二个单词，而年份是最后一个单词。

# reproducible example please!
d <- read.table(textConnection("
var1, var2
203, UPFRONT BROKERAGE FOR THE MONTH OF MARCH 2008
205, UPFRONT BROKERAGE FOR THE MONTH OF MAY 2010
206, UPFRONT BROKERAGE FOR THE MONTH OF MARCH 2008
207, UPFRONT BROKERAGE FOR THE MONTH OF APRIL 2009
204, UPFRONT BROKERAGE FOR THE MONTH OF MARCH 2008
"), header=TRUE, sep=",", stringsAsFactors=FALSE)

get_month <- function(s) {
  words <- unlist(strsplit(s, " "))
  words[length(words)-1]
}
get_year <- function(s) {
  words <- unlist(strsplit(s, " "))
  as.integer(words[length(words)])
}

d$month = sapply(d$var2, get_month)

d$year = lapply(d$var2, get_year)

d

产生所需的输出

> d
  var1                                           var2 month year
1  203  UPFRONT BROKERAGE FOR THE MONTH OF MARCH 2008 MARCH 2008
2  205    UPFRONT BROKERAGE FOR THE MONTH OF MAY 2010   MAY 2010
3  206  UPFRONT BROKERAGE FOR THE MONTH OF MARCH 2008 MARCH 2008
4  207  UPFRONT BROKERAGE FOR THE MONTH OF APRIL 2009 APRIL 2009
5  204  UPFRONT BROKERAGE FOR THE MONTH OF MARCH 2008 MARCH 2008