R警告:“项目,以取代的数目不替换长度的倍数”似乎不正确

问题描述:

我有一个为R脚本,看起来像这样:R警告:“项目,以取代的数目不替换长度的倍数”似乎不正确

# Create example data 
date <- c("11/09/2016", "11/02/2016", "11/16/2016", "11/23/2016") 
column_two <- c(4, 2, 3, 4) 
# Populate a data frame and make sure the dates have the correct class 
mydata <- data.frame(date, column_two) 
mydata$date <- strptime(mydata$date, format="%m/%d/%Y") 

print("The contents of mydata are:") 
print(mydata) 

# Create a dummy list (or vector, or array, or what is it?) 
foo <- rep(NA, 5) 
print("foo is initialized to:") 
print(foo) 
print("The class of foo is:") 
print(class(foo)) 

earlydate <- min(mydata$date) 
print(sprintf("Earliest date is: %s", earlydate)) 
print("The class of earlydate is:") 
print(class(earlydate)) 
print(sprintf("Length of earliest date is: %d", length(earlydate))) 
print(sprintf("Length of foo[2] is: %d", length(foo[2]))) 

# Attempt to set one variable equal to another 
foo[2] <- earlydate 

print("After assignment, foo looks like this:") 
print(foo) 
print("Now the classes of foo, foo[2], and foo[[2]] are:") 
print(class(foo)) 
print(class(foo[2])) 
print(class(foo[[2]])) 

从脚本打印输出如下所示:

> source("test_warning.R") 
[1] "The contents of mydata are:" 
     date column_two 
1 2016-11-09   4 
2 2016-11-02   2 
3 2016-11-16   3 
4 2016-11-23   4 
[1] "foo is initialized to:" 
[1] NA NA NA NA NA 
[1] "The class of foo is:" 
[1] "logical" 
[1] "Earliest date is: 2016-11-02" 
[1] "The class of earlydate is:" 
[1] "POSIXlt" "POSIXt" 
[1] "Length of earliest date is: 1" 
[1] "Length of foo[2] is: 1" 
[1] "After assignment, foo looks like this:" 
[[1]] 
[1] NA 

[[2]] 
[1] 0 

[[3]] 
[1] NA 

[[4]] 
[1] NA 

[[5]] 
[1] NA 

[1] "Now the classes of foo, foo[2], and foo[[2]] are:" 
[1] "list" 
[1] "list" 
[1] "numeric" 
Warning message: 
In foo[2] <- earlydate : 
    number of items to replace is not a multiple of replacement length 
> 

我有这么多的问题:

  • 为什么我得到一个警告,当foo[2]earlydate显然是相同的长度?
  • 为什么foo [2]的值设置为0而不是值为earlydate
  • 很明显,R会自动将foo变量及其元素强制为新的类;为什么这些类别中的任何一个(foofoo[2]foo[[2]])似乎都与earlydate变量的类别("POSIXlt" "POSIXt")相匹配?
  • 我该如何正确初始化foo,以便它成为一个向量,列表或类似数组的东西,其中包含能够支持这种类型值赋值的单独元素,而不会引发这种警告并导致这种反直觉行为?

那么,幕后的POSIXlt实际上是一个列表。

> class(unclass(earlydate)) 
[1] "list" 
> length(unclass(earlydate)) 
[1] 11 

赋值为0,因为这是列表的第一个元素;它是秒数,earlydate为0。

> unclass(earlydate)[1] 
$sec 
[1] 0 

我真的不知道为什么R不自动要挟foo可变进POSIXlt类;我的猜测是,对日期的强制通常很难。很明显,在fooNA的情况下要做什么,但如果其中一个元素已经是整数或字符串呢?但首先要自己强制使用,请使用as.POSIXlt

foo <- rep(as.POSIXlt(NA), 5) 

取决于你会用它做什么其他,你也可能会发现使用POSIXct是更好的解决方案,在这个答案讨论Integer to POSIXlt。它将日期存储为从原点开始的秒数,因此不是下面的列表,因此更适合于某些类型的计算和存储。