Question

我正在通过对另一列进行分组来处理列的滚动平均值。这些组是“PlayerName”，滚动平均值是“FDP”。这是我正在使用的测试数据：

PlayerName <- c("Player1", "Player1", "Player1", "Player1", "Player2", "Player2", "Player2", "Player3", "Player3", "Player3", "Player3", "Player3", "Player3")

Date <- c("2016-11-02", "2016-11-04", "2016-11-07", "2016-11-10", "2016-11-03", "2016-11-04", "2016-11-11", "2016-11-02", "2016-11-04", "2016-11-07", "2016-11-10", "2016-11-13", "2016-11-17")

FDP <- c(13, 12, 8, 20, 10, 11, 12, 35, 46, 24, 33, 40, 40)

DT <- data.table(PlayerName, Date, FDP)

我想添加第4列“RollAvg”，这是3日期滚动平均值。结果看起来像：

    PlayerName       Date FDP  RollAvg
 1:    Player1 2016-11-02  13    NA
 2:    Player1 2016-11-04  12    NA
 3:    Player1 2016-11-07   8    11
 4:    Player1 2016-11-10  20    13.3
 5:    Player2 2016-11-03  10    NA
 6:    Player2 2016-11-04  11    NA
 7:    Player2 2016-11-11  12    11
 8:    Player3 2016-11-02  35    NA
 9:    Player3 2016-11-04  46    NA
10:    Player3 2016-11-07  24    35
11:    Player3 2016-11-10  33    34.3
12:    Player3 2016-11-13  40    32.3
13:    Player3 2016-11-17  40    37.6

如果RollAvg列有NA或者省略它们，对我来说无关紧要。

我已经尝试了很多不同的方法来做到这一点，我无法得到任何工作。我知道还有其他类似的帖子，但我无法得到他们正在做的工作的结构。我觉得有一些简单的单行方式用sapply和rollmean做到这一点，但我无法弄明白。

感谢您的帮助！

Answer 1

正如其他几个SO posts建议的那样，您可以使用zoo包的rollapply()函数来解决此问题：

library(zoo)
library(dplyr)

PlayerName <- c("Player1", "Player1", "Player1", "Player1", "Player2", "Player2", "Player2", "Player3", "Player3", "Player3", "Player3", "Player3", "Player3")

Date <- c("2016-11-02", "2016-11-04", "2016-11-07", "2016-11-10", "2016-11-03", "2016-11-04", "2016-11-11", "2016-11-02", "2016-11-04", "2016-11-07", "2016-11-10", "2016-11-13", "2016-11-17")

FDP <- c(13, 12, 8, 20, 10, 11, 12, 35, 46, 24, 33, 40, 40)

data.frame(PlayerName, Date, FDP) %>% 
        mutate(Date = as.Date(Date)) %>% 
        arrange(PlayerName,Date) %>% 
        group_by(PlayerName) %>%
        mutate(RollAvg = zoo::rollapply(FDP,3,mean,by = 1,align = "right",fill = NA))

返回：

Source: local data frame [13 x 4]
Groups: PlayerName [3]

   PlayerName       Date   FDP  RollAvg
       <fctr>     <date> <dbl>    <dbl>
1     Player1 2016-11-02    13       NA
2     Player1 2016-11-04    12       NA
3     Player1 2016-11-07     8 11.00000
4     Player1 2016-11-10    20 13.33333
5     Player2 2016-11-03    10       NA
6     Player2 2016-11-04    11       NA
7     Player2 2016-11-11    12 11.00000
8     Player3 2016-11-02    35       NA
9     Player3 2016-11-04    46       NA
10    Player3 2016-11-07    24 35.00000
11    Player3 2016-11-10    33 34.33333
12    Player3 2016-11-13    40 32.33333
13    Player3 2016-11-17    40 37.66667

R编程 - 分组滚动平均值

1 个答案: