Question

我有一个名为df的数据框

dput(df)

structure(list(timestamp = structure(c(1L, 1L, 2L, 2L, 2L, 3L
), .Label = c("6/17/2016 11:58", "6/17/2016 12:00", "6/17/2016 15:30"
), class = "factor"), app = structure(c(2L, 2L, 2L, 2L, 1L, 1L
), .Label = c("db", "web"), class = "factor"), qmanager = structure(c(3L, 
3L, 3L, 4L, 1L, 2L), .Label = c("db01", "db02", "web01", "web04"
), class = "factor"), qname = structure(c(3L, 3L, 4L, 2L, 1L, 
1L), .Label = c("dbtest101", "test02", "test101", "test102"), class = "factor"), 
    que = c(500L, 600L, 66L, 12000L, 4000L, 666L), mas = c(15000L, 
    50000L, 15000L, 175000L, 5000L, 15000L)), .Names = c("timestamp", 
"app", "qmanager", "qname", "que", "mas"), class = "data.frame", row.names = c(NA, 
-6L))

我需要通过qname，qmanager和app

将这些数据最多聚合到5分钟

我正在尝试这个：

df$timestamp <- as.POSIXct(df$timestamp, format="%m/%d/%Y %H:%M")

library(xts)
df$timestamp<-align.time(df$timestamp, n=60*5)
df<-aggregate(que ~ cut(df$timestamp, "5 min"), df$qname,df$qmanager, df$app df[setdiff(names(df), "timestamp")], max)

无法让它发挥作用，任何想法？

Answer 1

这样的东西？

# install.packages('dplyr')    
library(dplyr)

df %>%
  mutate(min5_interval = as.POSIXct(timestamp, format = '%m/%d/%Y %H:%M') 
         %>% cut('5 min')) %>%
  group_by(min5_interval, qname, qmanager, app) %>%
  dplyr::summarise(max_que = max(que),
                   max_mas = max(mas))

Source: local data frame [5 x 6]
Groups: min5_interval, qname, qmanager [?]

        min5_interval     qname qmanager    app max_que max_mas
               (fctr)    (fctr)   (fctr) (fctr)   (int)   (int)
1 2016-06-17 11:58:00 dbtest101     db01     db    4000    5000
2 2016-06-17 11:58:00    test02    web04    web   12000  175000
3 2016-06-17 11:58:00   test101    web01    web     600   50000
4 2016-06-17 11:58:00   test102    web01    web      66   15000
5 2016-06-17 15:28:00 dbtest101     db02     db     666   15000

如何基于多个列聚合数据？

1 个答案: