部分 group by 语句以查找变量的多个实例的计数

Posted 2023-04-13

【中文标题】部分 group by 语句以查找变量的多个实例的计数

【英文标题】：Partial group by statements to find the count for multiple instances of a variable

【发布时间】：2015-02-09 22:33:46

【问题描述】：

我正在为一项相当简单的任务而苦苦挣扎。我有以下数据，并想为每个 visit_high 查找 event_list 中的项目计数。所以它可能如下所示。

Visit_high   visit    event_list
101            1          3
101            2          5
102            1          2
103            1          6
103            2          8
103            3          5
...

Visit high 是用户 id，visit 是指他们的访问次数，event list 是他们采取的操作数。因此，用户 101 两次访问该网站，并在第一次访问时执行了 3 次操作，在第二次访问时执行了 5 次操作。

> dput(tail(mydf[1:50,c(5,10)], 10))
structure(list(event_list = structure(c(2L, 2L, 2L, 2L, 76L, 
36L, 64L, 37L, 14L, 25L), .Label = c("", "100,101,102,115,116", 
"100,101,102,115,116,146", "100,101,102,116", "100,101,102,116,146", 
"100,101,115,116", "100,101,117,118", "100,102,115,116", "100,102,115,116,146", 
"100,102,116", "100,102,116,146", "100,107,115,116", "100,107,116", 
"100,115,116", "100,115,116,146", "100,116", "100,116,146", "100,117", 
"102,115,116", "102,115,116,146", "102,116", "102,116,146", "107,115,116", 
"108,117,118", "115,116", "115,116,146", "116", "116,146", "202", 
"202,120", "205,100,101,109,117,118", "206,115,116", "206,115,116,146", 
"206,116", "206,116,146", "206,214,115,116", "206,214,115,116,146", 
"206,214,116", "206,214,116,146", "206,215,115,116", "206,215,115,116,146", 
"207,102,115,116", "207,102,115,116,146", "207,102,116", "207,102,116,146", 
"207,115,116", "208,100,101,102,115,116", "208,100,101,102,116", 
"208,100,102,115,116", "208,100,115,116", "208,102,109,115,116", 
"208,102,109,116", "208,102,115,116", "208,102,116", "208,109,115,116", 
"208,109,115,116,146", "208,109,116", "208,115,116", "208,116", 
"210,102,108,115,116", "210,102,108,116", "212,102,109,115,116", 
"212,102,109,116", "212,109,115,116", "212,109,116", "212,115,116", 
"214,100,101,102,115,116", "214,100,101,102,115,116,146", "214,100,115,116", 
"214,100,115,116,146", "214,100,116", "214,100,116,146", "214,102,115,116", 
"214,102,115,116,146", "214,102,116", "214,115,116", "214,115,116,146", 
"214,116", "214,116,146", "214,207,102,115,116", "214,221,102,115,116", 
"214,221,102,115,116,146", "215,100,101,102,115,116", "215,100,101,102,115,116,146", 
"215,100,101,102,116", "215,100,101,115,116", "215,100,102,115,116", 
"215,100,102,116", "215,100,115,116", "215,100,115,116,146", 
"215,100,116", "215,102,115,116", "215,102,115,116,146", "215,102,116", 
"215,115,116", "215,115,116,146", "215,116", "215,207,102,115,116", 
"215,207,102,116", "215,221,100,102,115,116", "215,221,100,102,116", 
"215,221,102,115,116", "215,221,102,116", "220,102,115,116", 
"221,100,102,115,116", "221,100,102,115,116,146", "221,100,102,116", 
"221,102,115,116", "221,102,115,116,146", "221,102,116", "226,100,117,119,120", 
"227,102,115,116", "227,102,116", "228,102,115,116", "232,102,115,116", 
"234,102,115,116", "235"), class = "factor"), visid_high = c(2710815361820866560, 
2710815518587167232, 2710815707565725184, 2710815726893081600, 
2710815857889578496, 2710815857889578496, 2710815857889578496, 
2710815883659387904, 2710815902986739712, 2710815950231374336
)), .Names = c("event_list", "visid_high"), row.names = 41:50, class = "data.frame")

我有每个访问者 ID 的访问次数，但我对如何区分每个 visit_high 实例有点迷茫。

event_sum = cbind(mmf$visid_high, mmf$event_list, sapply(strsplit(mmf$event_list, ","), length))

【问题讨论】：

您到底想达到什么目的？每个Visit_high 的event_list 的总和？例如：访客 101 的 8 个操作？

如果您的示例使用一致的拼写，例如visid 与 visit。但我们真正需要的是您提供的非常短的样本输入的样本输出。

【参考方案1】：

希望我的问题是正确的（调用您的数据 DF）：

myfun <- function(row)
  data.frame(event_list = unlist(strsplit(row[1],",")), visid_high = row[2])

table(do.call(rbind, apply(DF,1,myfun)))

          visid_high
event_list 2.710815e+18 2.710816e+18
       100            1            4
       101            1            3
       102            1            3
       115            1            9
       116            1            9
       214            0            3
       206            0            2
       109            0            1
       212            0            1
       146            0            1

【参考方案2】：

这个怎么样。

# dummy data based on your example
du <- data.frame(Visit_high=c(101,101,102,103,103,103),  
                 visit=c(1,2,1,1,2,3), 
                 event_list=c(3,5,2,6,8,5))
# a function to sum all visits for a given visit-high
fu <- function(vh) 
  rows_for_this_user <- which(du$Visit_high==vh)
  events_for_this_user <- sum(du$event_list[rows_for_this_user]) 
  #return a vector with the user id and events count
  c(Visit_high=vh, event_sum=events_for_this_user)


data.frame(t(sapply(unique(du$Visit_high), fu)))

样本输出。

      Visit_high event_sum
1        101         8
2        102         2
3        103        19

【参考方案3】：

如果您需要group 的sum，有很多方法可以做到。

library(data.table)
setDT(du)[, list(event_sum=sum(event_list)), by=Visit_high]
#   Visit_high event_sum
#1:        101         8
#2:        102         2
#3:        103        19

或base R 方法

 aggregate(cbind(event_sum=event_list)~Visit_high, du, FUN=sum)

数据

du <- data.frame(Visit_high=c(101,101,102,103,103,103),  
             visit=c(1,2,1,1,2,3), 
             event_list=c(3,5,2,6,8,5))