在R中提取列值从一个日期更改为下一个日期的行

Question

晚上好！我有以下data.frame，我想从中提取限定符从一个月到下个月更改的行。不幸的是，我不知道如何解决这个问题......

数据：

id <- c("A","B","C","D","A","B","C","D","A","B","C","D")
qualifier <- c("G","H","I","I","I","H","Y","I","I","I","I","I")
date <- c("01/31/15","01/31/15",'01/31/15','01/31/15','02/28/15', "02/28/15", "02/28/15","02/28/15","03/31/15","03/31/15","03/31/15","03/31/15")
sample.data <- data.frame(id,qualifier,date)

“id”对于任何给定日期都是唯一的，但在其他日期可能存在也可能不存在。每个“id”都有一个限定符“I”或“H”。所有日期均为月末。我想提取从一个月到下个月 id 限定符从“I”或“G”变为“H”或“Y”的所有行。因此，要指定，只有从 01/31/15 到 02/28/15 的变化是相关的，而不是从 01/31/15 到 03/31/15 的变化。鉴于上面的 data.frame，我感兴趣的更改将是第 7 行，当 "id" C 的限定符从 "I" （从第 3 行）更改为 "Y" 时。理想情况下，提取的第 7 行也将具有第 3 行的限定符的值。

结果应该是这样的：

id   qualifier  date       qualifier.previous.month
C    H          02/28/15   I

提前感谢您的帮助！

Answer 1

# Change the date to a Date format
sample.data$date <- as.Date(sample.data$date, format = "%m/%d/%y")

# Collect the month value from the date
sample.data$month <- as.numeric(format(sample.data$date, "%m")) 

# Create a copy of the sample dataframe
sample.data_shifted <- sample.data

# Shift the months by one in the copied dataframe
sample.data_shifted$month <- sample.data_shifted$month + 1

# Merge the dataframes, so each ID/month row has the current and previous month qualifier 
sample.data_merged <- merge(
  sample.data_shifted, sample.data, 
  by = c("id", "month"), 
  suffixes = c(".previous.month", ".current.month")
  )

# Tidying up dates
sample.data_merged$date <- sample.data_merged$date.current.month

sample.data_merged <- subset(sample.data_merged, select = -c(date.previous.month, date.current.month))

# We create a logical vector telling us which qualifiers changed between which months
previous_month_ig <- sample.data_merged$qualifier.previous.month %in% c("I", "G")
current_month_hy <- sample.data_merged$qualifier.current.month %in% c("H", "Y")

# Now we only look at IDs/months where the qualifier changed from I/G to H/Y
sample.data_final <- sample.data_merged[previous_month_ig & current_month_hy, ]

sample.data_final

  id month qualifier.previous.month qualifier.current.month       date
5  C     2                        I                       Y 2015-02-28

这应该符合您的规范 - 让它更清楚有点冗长。这里的想法是合并回您的数据框，月份偏移一。如果您每个月只有一个日期，这应该可以工作。

Answer 2

在transform 中使用ave 的基本R 选项可能会有所帮助

transform(
  sample.data,
  qualifier.previous.month = ave(qualifier, id, FUN = function(x) c(NA, x[-length(x)]))
)

给了

   id qualifier     date qualifier.previous.month
1   A         G 01/31/15                     <NA>
2   B         H 01/31/15                     <NA>
3   C         I 01/31/15                     <NA>
4   D         I 01/31/15                     <NA>
5   A         I 02/28/15                        G
6   B         H 02/28/15                        H
7   C         Y 02/28/15                        I
8   D         I 02/28/15                        I
9   A         I 03/31/15                        I
10  B         I 03/31/15                        H
11  C         I 03/31/15                        Y
12  D         I 03/31/15                        I

---

使用shift 的data.table 选项也可以实现

> setDT(sample.data)[, qualifier.previous.month := shift(qualifier), id][]
    id qualifier     date qualifier.previous.month
 1:  A         G 01/31/15                     <NA>
 2:  B         H 01/31/15                     <NA>
 3:  C         I 01/31/15                     <NA>
 4:  D         I 01/31/15                     <NA>
 5:  A         I 02/28/15                        G
 6:  B         H 02/28/15                        H
 7:  C         Y 02/28/15                        I
 8:  D         I 02/28/15                        I
 9:  A         I 03/31/15                        I
10:  B         I 03/31/15                        H
11:  C         I 03/31/15                        Y
12:  D         I 03/31/15                        I

Answer 3

library(data.table)

id <- c("A","B","C","D","A","B","C","D","A","B","C","D")
qualifier <- c("G","H","I","I","I","H","Y","I","I","I","I","I")
date <- c("01/31/15","01/31/15",'01/31/15','01/31/15','02/28/15', "02/28/15", "02/28/15","02/28/15","03/31/15","03/31/15","03/31/15","03/31/15")
sample.data <- data.frame(id,qualifier,date)

setDT(sample.data)

sample.data[, qualifier.previous.month := shift(.SD, 1, 0, "lag"), by = id]
> sample.data
    id qualifier     date qualifier.previous.month
 1:  A         G 01/31/15                        0
 2:  B         H 01/31/15                        0
 3:  C         I 01/31/15                        0
 4:  D         I 01/31/15                        0
 5:  A         I 02/28/15                        G
 6:  B         H 02/28/15                        H
 7:  C         Y 02/28/15                        I
 8:  D         I 02/28/15                        I
 9:  A         I 03/31/15                        I
10:  B         I 03/31/15                        H
11:  C         I 03/31/15                        Y
12:  D         I 03/31/15                        I