r - 如何计算每三个连续元素的平均值

Question

这是一个列表。

 1 2 3 4 5 6 7 8 9 10. 

我想计算每三个连续元素的平均值。例如，输出将是 NA NA 2 3 4 5 6 7 8 9。

这个怎么做？

问候

Answer 1

您可以?embed尝试?rowMeans：

v <- 1:10

m <- embed(v, 3)
m
#    [,1] [,2] [,3]
#[1,]    3    2    1
#[2,]    4    3    2
#[3,]    5    4    3
#[4,]    6    5    4
#[5,]    7    6    5
#[6,]    8    7    6
#[7,]    9    8    7
#[8,]   10    9    8

rowMeans(m)
# 2 3 4 5 6 7 8 9

编辑：另一个解决方案是?filter：

filter(x=v, filter=rep(1/3, 3), sides=1)
# Time Series:
# Start = 1 
# End = 10 
# Frequency = 1 
#  [1] NA NA  2  3  4  5  6  7  8  9

Answer 2

你甚至可以使用rollapply 或者rollmean来自 zoo 包

> library(zoo)
> v <- 1:10
> rollapply(v, width=3, align="right", FUN=mean, fill=NA )
 [1] NA NA  2  3  4  5  6  7  8  9

> rollmean(v, k=3, align="right",  fill=NA )
 [1] NA NA  2  3  4  5  6  7  8  9

Answer 3

This is fourth way, using the lag function:

v <- 1:10
rowMeans(do.call(cbind, lapply(0:2, lag, x=as.ts(v))))
# [1] NA NA  2  3  4  5  6  7  8  9 NA NA

You can wrap this in na.omit to remove the NAs.

Benchmarks

library(microbenchmark)
library(zoo)
v <- 1:10000
f.embed <- function() rowMeans(embed(v, 3))
f.filter <- function() filter(x=v, filter=rep(1/3, 3), sides=1)
f.lag <- function() rowMeans(do.call(cbind, lapply(0:2, lag, x=as.ts(v))))
f.rollmean <-function() rollapply(v, width=3, align="right", FUN=mean, fill=NA ) 
microbenchmark(f.embed(), f.filter(), f.lag(), f.rollmean())
# Unit: microseconds
#          expr    min     lq median     uq     max neval
#     f.embed()  486.7  499.8  505.6  517.1  1633.1   100
#    f.filter()  285.3  300.7  307.2  316.6   912.5   100
#       f.lag() 1601.6 1640.9 1677.0 2188.3  2838.7   100
#  f.rollmean() 4265.4 4853.5 4902.0 5364.8 52098.9   100