日期上的嵌套 If 语句答案

【问题标题】：nested If statement on dates日期上的嵌套 If 语句
【发布时间】：2017-01-31 20:09:37
【问题描述】：

我有一个数据框df，如下所示。

Id     ProcessDate
10     2011-12-29 14:14:00
11     2011-12-29 14:16:00
12     2011-12-29 14:14:00
13     2011-12-29 14:20:00
14     2011-12-29 14:49:00
15     2011-12-29 14:51:00
16     2011-12-29 14:53:00
17     2011-12-29 15:11:00
18     2011-12-29 15:13:00 
19     2011-12-29 15:10:00
20     2011-12-29 15:21:00
21     2011-12-29 14:34:00
22     2011-12-29 15:26:00

我正在尝试创建第三列 Status，它将包含这三个值之一 {Before, during , after } 基于此条件。

 if  (df$ProcessDate < 2011-12-29 14:48:00)
 then  df$Status = "Before"
 else if (df$ProcessDate > 2011-12-29 14:48:00 & df$ProcessDate < 2011-12-29 15:16:00)
 then  df$Status = "Between"
 else  df$Status = "After"

最终的数据框应如下所示。

Id     ProcessDate              Status
10     2011-12-29 14:14:00      Before
11     2011-12-29 14:16:00      Before
12     2011-12-29 14:14:00      Before
13     2011-12-29 14:20:00      Before
14     2011-12-29 14:49:00      Between
15     2011-12-29 14:51:00      Between       
16     2011-12-29 14:53:00      Between
17     2011-12-29 15:11:00      Between
18     2011-12-29 15:13:00      Between
19     2011-12-29 15:10:00      Between
20     2011-12-29 15:21:00      After
21     2011-12-29 14:34:00      After
22     2011-12-29 15:26:00      After

我尝试了一些方法，但没有成功，非常感谢您对此问题的任何帮助。

【问题讨论】：

有a few examples 的similar questions - 你尝试了什么，为什么没有成功？
@KimJenkins 我认为您的倒数第二行的状态应该是“之前”，对吧？

标签： r if-statement dataframe

【解决方案1】：

子集赋值

对于这种特殊情况，在基本 R 中执行此操作的一个非常简单的方法是将所有内容设置为 'Between'，然后使用子集分配来更改应该是其他内容的行：

df$ProcessDate <- as.POSIXct(df$ProcessDate)    # skip if already parsed to datetime

df$Status <- 'Between'
df$Status[df$ProcessDate < as.POSIXct('2011-12-29 14:48:00')] <- 'Before'
df$Status[df$ProcessDate >= as.POSIXct('2011-12-29 15:16:00')] <- 'After'

df
##    Id         ProcessDate  Status
## 1  10 2011-12-29 14:14:00  Before
## 2  11 2011-12-29 14:16:00  Before
## 3  12 2011-12-29 14:14:00  Before
## 4  13 2011-12-29 14:20:00  Before
## 5  14 2011-12-29 14:49:00 Between
## 6  15 2011-12-29 14:51:00 Between
## 7  16 2011-12-29 14:53:00 Between
## 8  17 2011-12-29 15:11:00 Between
## 9  18 2011-12-29 15:13:00 Between
## 10 19 2011-12-29 15:10:00 Between
## 11 20 2011-12-29 15:21:00   After
## 12 21 2011-12-29 14:34:00  Before
## 13 22 2011-12-29 15:26:00   After

`cut`

专门设计的方法是使用cut，它有一个cut.POSIXt 方法。除了您已经想要的数据之外，它还需要在数据之前和之后设置断点，但对分类数据来说是一个很好的因素。

df$Status <- cut(df$ProcessDate, 
                 breaks = c(min(df$ProcessDate), 
                          as.POSIXct(c('2011-12-29 14:48:00', '2011-12-29 15:16:00')), 
                          max(df$ProcessDate) + 1), 
                 labels = c('Before', 'Between', 'After'))

嵌套的`ifelse` 调用

最常见和通用的基本版本是嵌套的ifelse 调用，它看起来很难看（特别是如果它们很多的话），但评估速度很快，因为ifelse 是矢量化的，而if 不是：

df$Status <- ifelse(df$ProcessDate < as.POSIXct('2011-12-29 14:48:00'), 
                    'Before', 
                    ifelse(df$ProcessDate < as.POSIXct('2011-12-29 15:16:00'), 
                           'Between', 
                           'After'))

dplyr

dplyr::case_when 是嵌套 ifelse 调用的一个很好的替代方案。它依次评估每个条件并返回相应的值：

library(dplyr)

df %>% mutate(
    ProcessDate = as.POSIXct(ProcessDate),    # skip this line if already datetime
                       # if this is true,                      then return "Before"
    Status = case_when(.$ProcessDate < as.POSIXct('2011-12-29 14:48:00') ~ 'Before',
                       # for the rest, if this is true,             return "Between"
                       .$ProcessDate < as.POSIXct('2011-12-29 15:16:00') ~ 'Between',
                       # always true, so make the rest "After"
                       TRUE ~ 'After'))

所有版本都返回相同的东西，除了 cut，它返回一个因子而不是字符向量。

【讨论】：

【解决方案2】：

试试这个：

left <- as.POSIXct("12/29/2011 14:48", format = "%m/%d/%Y %H:%M") 
right <- as.POSIXct("12/29/2011 15:16", format = "%m/%d/%Y %H:%M") 
DT[, Status := ifelse(ProcessDate < left, "before", 
            ifelse(ProcessDate > right, "after", "between"))]

它给出：

    Id         ProcessDate  Status
 1: 10 2011-12-29 14:14:00  before
 2: 11 2011-12-29 14:16:00  before
 3: 12 2011-12-29 14:14:00  before
 4: 13 2011-12-29 14:20:00  before
 5: 14 2011-12-29 14:49:00 between
 6: 15 2011-12-29 14:51:00 between
 7: 16 2011-12-29 14:53:00 between
 8: 17 2011-12-29 15:11:00 between
 9: 18 2011-12-29 15:13:00 between
10: 19 2011-12-29 15:10:00 between
11: 20 2011-12-29 15:21:00   after
12: 21 2011-12-29 15:34:00   after
13: 22 2011-12-29 15:26:00   after

结果与上面相同，可矢量化ifelse() 和data.table

【讨论】：

【解决方案3】：

这可能是一种可能的解决方案

ids = c(10, 11, 12, 13, 14, 15, 16, 17, 18,  19, 20, 21, 22)      
date = c('2011-12-29 14:14:00', '2011-12-29 14:16:00', '2011-12-29 14:14:00', '2011-12-29 14:20:00', '2011-12-29 14:49:00', '2011-12-29 14:51:00', '2011-12-29 14:53:00', '2011-12-29 15:11:00', '2011-12-29 15:13:00', '2011-12-29 15:10:00', '2011-12-29 15:21:00', '2011-12-29 14:34:00', '2011-12-29 15:26:00')
df <- data.frame(Id = ids, 
                 ProcessDate = strptime(date, format = '%Y-%m-%d %H:%M:%S'))


date.status.before <- strptime('2011-12-29 14:48:00', format = '%Y-%m-%d %H:%M:%S')
date.status.after <- strptime('2011-12-29 15:16:00', format = '%Y-%m-%d %H:%M:%S')
ProcessDateStatus <- function(process.date) {
  if  (process.date < date.status.before)
    "Before"
  else if (process.date > date.status.before & process.date < date.status.after)
    "Between"
  else 
    "After"  
}
df$Status <- lapply(df$ProcessDate, ProcessDateStatus)

导致

   Id         ProcessDate  Status
1  10 2011-12-29 14:14:00  Before
2  11 2011-12-29 14:16:00  Before
3  12 2011-12-29 14:14:00  Before
4  13 2011-12-29 14:20:00  Before
5  14 2011-12-29 14:49:00 Between
6  15 2011-12-29 14:51:00 Between
7  16 2011-12-29 14:53:00 Between
8  17 2011-12-29 15:11:00 Between
9  18 2011-12-29 15:13:00 Between
10 19 2011-12-29 15:10:00 Between
11 20 2011-12-29 15:21:00   After
12 21 2011-12-29 14:34:00  Before
13 22 2011-12-29 15:26:00   After

【讨论】：

【解决方案4】：

一种可能的解决方案是将您的时间转换为纪元值，然后进行比较。这可以通过使用 as.integer(as.POSIXct("Time")) 来完成，如下所示

df = NULL
df$ids = c(10, 11, 12, 13, 14, 15, 16, 17, 18,  19, 20, 21, 22)      
df$date = c('2011-12-29 14:14:00', '2011-12-29 14:16:00', '2011-12-29      14:14:00', '2011-12-29 14:20:00', '2011-12-29 14:49:00', '2011-12-29 14:51:00', '2011-12-29 14:53:00', '2011-12-29 15:11:00', '2011-12-29 15:13:00', '2011-12-29 15:10:00', '2011-12-29 15:21:00', '2011-12-29 14:34:00', '2011-12-29 15:26:00')
df = as.data.frame(df)
df$date = as.integer(as.POSIXct(df$date))

upper   = as.integer(as.POSIXct('2011-12-29 15:16:00'))
lower   = as.integer(as.POSIXct('2011-12-29 14:48:00'))

您将拥有如下转换后的日期列

> df
    ids       date
1   10 1325148240
2   11 1325148360
3   12 1325148240
4   13 1325148600
5   14 1325150340
6   15 1325150460
7   16 1325150580
8   17 1325151660
9   18 1325151780
10  19 1325151600
11  20 1325152260
12  21 1325149440
13  22 1325152560

然后你可以简单地进行数字比较

for(i in c(1:nrow(df))){
    if(df$date[i] < lower)
            df$Status[i] = "Before"
    else if(df$date[i] > lower & df$date[i] < upper)
            df$Status[i] = "Between"
    else
            df$Status[i] = "After"
}

导致输出

> df
    ids       date  Status
1   10 1325148240  Before
2   11 1325148360  Before
3   12 1325148240  Before
4   13 1325148600  Before
5   14 1325150340 Between
6   15 1325150460 Between
7   16 1325150580 Between
8   17 1325151660 Between
9   18 1325151780 Between
10  19 1325151600 Between
11  20 1325152260   After
12  21 1325149440  Before
13  22 1325152560   After

【讨论】：

要保留日期的格式，您可以在每行的条件内执行从日期和时间到纪元的转换。
转换为数字并没有真正的意义；比较运算符（< 等）直接在 POSIX*t 日期时间上工作，因此您所做的只是让数据更难阅读。
根据我之前的评论，您可以保留数据格式。转换不需要是明确的。进行转换是为了在比较日期时显示纪元的概念。

子集赋值

cut

嵌套的ifelse 调用

dplyr

`cut`

嵌套的`ifelse` 调用