根据时间间隔组合行值答案

【问题标题】：Combine the row values based on time interval根据时间间隔组合行值
【发布时间】：2021-03-08 10:43:55
【问题描述】：

我有一个这样的数据框

node <- c("ABCC","ABCC","ABCC","ABCC","ABCC","ABCC","ABCC","ABCC","ABCC")
activity <-c("NODE_ISOLATION","NE_NOT_REACH_VIA_PRIMARY_MNG_INTERF","NODE_ISOLATION","LOSS_OF_MULTIPLEX_SECTION-OMS_A","NODE_ISOLATION","NE_NOT_REACH_VIA_PRIMARY_MNG_INTERF","NE_NOT_REACH_VIA_PRIMARY_MNG_INTERF", "UNDERLYING_RESOURCE_UNAVAILABLE-OMS_A","UNDERLYING_RESOURCE_UNAVAILABLE-OMS_A") 

e <-c("2020-05-08 16:11:58","2020-05-08 16:11:58","2020-05-08 16:30:07","2020-05-09 03:00:08","2020-05-09 03:08:08","2020-05-09 03:28:08","2020-05-09 13:08:08","2020-05-09 13:10:08","2020-05-09 13:28:08")

df <- data.frame(node, activity, e)
df

我想根据 30 分钟的时间间隔组合行值。所需的输出如下

node <- c("ABCC","ABCC","ABCC")

activity <-c("NODE_ISOLATION,NE_NOT_REACH_VIA_PRIMARY_MNG_INTERF,NODE_ISOLATION","LOSS_OF_MULTIPLEX_SECTION-OMS_A,NODE_ISOLATION,NE_NOT_REACH_VIA_PRIMARY_MNG_INTERF","NE_NOT_REACH_VIA_PRIMARY_MNG_INTERF,UNDERLYING_RESOURCE_UNAVAILABLE-OMS_A,UNDERLYING_RESOURCE_UNAVAILABLE-OMS_A") 

cnt <- c(3,3,3)
df1 <- data.frame(node, activity, cnt)
df1

【问题讨论】：

标签： r dplyr

【解决方案1】：

这看起来像是时间差的累积总和，带有重置。

设置threshold，这是组中的最大时差。在这种情况下，使用 30 分钟（30 * 60 = 1800 秒）。

确保e 是POSIXct。

函数cumsum_with_reset 将计算时间差的累积总和，一旦超过threshold，下一组重置为零。

因此，使用此自定义函数的方法是：首先，group_by(node) 用于在每个 node 内进行的评估。计算行之间的差异diff。使用自定义函数确定每个不超过 30 分钟的group。然后，group_by 这个新的group，计算每个组的行数，并使用toString 将activity 放入集体逗号分隔值中。

library(tidyverse)

threshold <- 30 * 60 # 30 minutes * 60 seconds

df$e <- as.POSIXct(df$e)

cumsum_with_reset <- function(x, threshold) {
  cumsum <- 0
  group <- 0
  result <- numeric()
  for (i in seq_along(x)) {
    cumsum <- cumsum + x[i]
    if (cumsum >= threshold) {
      group <- group + 1
      cumsum <- 0
    }
    result = c(result, group)
  }
  return (result)
}

df %>%
  group_by(node) %>%
  mutate(diff = c(0, diff(e)),
         group = cumsum_with_reset(diff, threshold)) %>%
  group_by(group, .add = TRUE) %>%
  summarise(cnt = n(),
            activity = toString(activity)) %>%
  dplyr::select(-group)

输出

  node    cnt activity                                                                                                         
  <chr> <int> <chr>                                                                                                            
1 ABCC      3 NODE_ISOLATION, NE_NOT_REACH_VIA_PRIMARY_MNG_INTERF, NODE_ISOLATION                                              
2 ABCC      3 LOSS_OF_MULTIPLEX_SECTION-OMS_A, NODE_ISOLATION, NE_NOT_REACH_VIA_PRIMARY_MNG_INTERF                             
3 ABCC      3 NE_NOT_REACH_VIA_PRIMARY_MNG_INTERF, UNDERLYING_RESOURCE_UNAVAILABLE-OMS_A, UNDERLYING_RESOURCE_UNAVAILABLE-OMS_A

【讨论】：

非常感谢您的贡献。我已经达到了期望的输出。
我有一个问题是任何两个活动同时发生如何修改？
您能进一步澄清一下吗？也许用例子更新你的问题？如果同时进行 2 个活动，您希望发生什么？
您的示例的前两行似乎具有相同的时间。但看起来我的输出与您想要的数据框df1 相同。你想要不同的结果吗？如果是这样，请随时将您的问题编辑为您最终的df1 应该是什么样子，并带有同时具有活动的示例数据。我很乐意回来再次编辑答案。

【解决方案2】：

node <- c("ABCC","ABCC","ABCC","ABCC","ABCC","ABCC","ABCC","ABCC","ABCC")
activity <-c("NODE_ISOLATION","NE_NOT_REACH_VIA_PRIMARY_MNG_INTERF","NODE_ISOLATION","LOSS_OF_MULTIPLEX_SECTION-OMS_A","NODE_ISOLATION","NE_NOT_REACH_VIA_PRIMARY_MNG_INTERF","NE_NOT_REACH_VIA_PRIMARY_MNG_INTERF", "UNDERLYING_RESOURCE_UNAVAILABLE-OMS_A","UNDERLYING_RESOURCE_UNAVAILABLE-OMS_A") 

e <-c("2020-05-08 16:11:58","2020-05-08 16:11:58","2020-05-08 16:30:07","2020-05-09 03:00:08","2020-05-09 03:08:08","2020-05-09 03:28:08","2020-05-09 13:08:08","2020-05-09 13:10:08","2020-05-09 13:28:08")

df <- data.frame(node, activity, e)

library(dplyr)
##### Change e format 
df <- df %>% mutate(e=as.POSIXct(e))

##### 1st Breaks option automatic 
start <- as.POSIXct("2020-05-08 16:00:00") # Can be min(as.POSIXct(df$e))
end <- max(as.POSIXct(df$e))+30*60
breaks_a <- seq(start,end,"30 min")

##### 2nd Breaks option manual
breaks_m <- c(as.POSIXct("2020-05-08 16:00:00"),
              as.POSIXct("2020-05-08 16:31:00"),
              as.POSIXct("2020-05-09 03:30:00"),
              as.POSIXct("2020-05-09 13:30:00"))

df <- df %>% mutate(e_cut_automatic=cut(e,breaks_a),
                    e_cut_manual=cut(e,breaks_m))

count_act <- function(x) data.frame(node=x$node[1],
                                    activity=paste(x$activity,collapse=","),
                                    cnt=nrow(x))

count_n <- function(x) do.call("rbind",lapply(split(x,x$node),count_act))


##### With automatic breaks (recomend)
r <- do.call("rbind",lapply(split(df,df$e_cut_automatic),count_n))
print(r,row.names = FALSE)

输出：

##### With manual breaks
r <- do.call("rbind",lapply(split(df,df$e_cut_manual),count_n))
print(r,row.names = FALSE)

输出：

【讨论】：

大数据集有什么建议
当然，看看我的新答案。

【解决方案3】：

试试这个：

node <- c("ABCC","ABCC","ABCC","ABCC","ABCC","ABCC","ABCC","ABCC","ABCC")
activity <-c("NODE_ISOLATION","NE_NOT_REACH_VIA_PRIMARY_MNG_INTERF","NODE_ISOLATION","LOSS_OF_MULTIPLEX_SECTION-OMS_A","NODE_ISOLATION","NE_NOT_REACH_VIA_PRIMARY_MNG_INTERF","NE_NOT_REACH_VIA_PRIMARY_MNG_INTERF", "UNDERLYING_RESOURCE_UNAVAILABLE-OMS_A","UNDERLYING_RESOURCE_UNAVAILABLE-OMS_A") 
e <-c("2020-05-08 16:11:58","2020-05-08 16:11:58","2020-05-08 16:30:07","2020-05-09 03:00:08","2020-05-09 03:08:08","2020-05-09 03:28:08","2020-05-09 13:08:08","2020-05-09 13:10:08","2020-05-09 13:28:08")

df <- data.frame(node, activity, e)

as.data.table(df) %>% mutate(e_cut=cut(as.POSIXct(e),"30 min")) %>%
  group_by(node, e_cut) %>%
  summarise(activity=paste(activity,collapse=","),cnt = n())%>%
  select(-e_cut)

【讨论】：