R在比较多个条件后从重复行中选择一行答案

【问题标题】：R select one row from duplicated rows after compare multi conditionsR在比较多个条件后从重复行中选择一行
【发布时间】：2015-09-18 00:36:02
【问题描述】：

我从大量数据中获得了这些重复的记录。现在，我需要从这些重复的行中选择一行。

ID <- c("6820","6820","17413","17413","38553","38553","52760","52760","717841","717841","717841","747187","747187","747187")
date <- c("2014-06-12","2015-06-11","2014-05-01","2014-05-01","2014-06-12","2015-06-11","2014-10-24","2014-10-24","2014-05-01","2014-05-01","2014-12-02","2014-03-01","2014-05-12","2014-05-12")
type <- c("ST","ST","MC","MC","LC","LC","YA","YA","YA","YA","MC","LC","LC","MC")
level <-c("firsttime","new","new","active","active","active","firsttime","new","active","new","active","new","active","active")
data <- data.frame(ID,date,type,level)

数据框将如下所示：

我想比较一下：对于每个 ID，如果它们的日期不同，则将它们全部保存在 df.right 中；如果日期相同，则比较类型，按照LC>MC>YA>ST的顺序选择它们（例如选择MC而不是YA），将它们放入df.right；如果类型相同，则比较级别，按活动>新>第一次的顺序选择它们（例如，第一次选择新的），然后将选择的放入df.right。

我尝试使用 foreach，这只是第一步，它不适用于 ID 有 3 个重复行。

foreach (i=unique(data$ID), .combine='rbind') %do% {data[data$ID==i, "date"][1] == data[data$ID==i, "date"][2])
b <- data[data$ID==i,]}

结果应该是这样的：有人知道该怎么做吗？真的很感激。谢谢

【问题讨论】：

标签： r

【解决方案1】：

dplyr 包很适合这种事情

使用因子，您可以指定您希望类别排序的方式。然后，您可以为每个唯一 ID/日期对选择每种类型和级别中的第一个。

library(dplyr)

ID <- c("6820","6820","17413","17413","38553","38553","52760","52760","717841","717841","717841","747187","747187","747187")
date <- c("2014-06-12","2015-06-11","2014-05-01","2014-05-01","2014-06-12","2015-06-11","2014-10-24","2014-10-24","2014-05-01","2014-05-01","2014-12-02","2014-03-01","2014-05-12","2014-05-12")
type <- c("ST","ST","MC","MC","LC","LC","YA","YA","YA","YA","MC","LC","LC","MC")
level <-c("firsttime","new","new","active","active","active","firsttime","new","active","new","active","new","active","active")

type <- factor(type, levels=c("LC", "MC", "YA", "ST"))

level <- factor(level, levels=c("active", "new", "firsttime"))

data <- data.frame(ID,date,type,level)

df.right <- data %>%
  group_by(ID, date) %>%
  filter(type == sort(type)[1]) %>%
  filter(level == sort(level)[1])

【讨论】：

我不认为这个答案会产生正确的输出
@pcantalupo 它与 OP 的样本输出不完全匹配，但我认为 OP 的样本输出不正确，因为在第 13 行和第 14 行之间，应该保留第 13 行（而不是 14），因为 LC优先于 MC
嗯，我想知道为什么事情看起来不太对劲。
这很优雅；我建议的改进是使用arrange 按类型和级别排序，然后使用top_n 拉出顶部元素；所以在group_by之后，就是arrange(type, level) %>% top_n(1)

【解决方案2】：

这里的诀窍是根据需要对type 和level 的级别进行排序。然后需要进行两次重复数据删除：首先，根据ID, date, type 列删除重复行，其次，根据前两列删除重复行：

type = factor(type, levels=c("ST","YA","MC","LC"))
level = factor(level, levels=c("active","new","firsttime"))
data <- data.frame(ID,date,type,level)

d = with(data, data[order(ID, date, type, level),])
e = d[-which(duplicated(d[,1:3])),]
df.right = e[-which(duplicated(e[,1:2])),]
df.right = df.right[order(as.numeric(as.character(df.right$ID)), df.right$date),]
df.right

输出：

       ID       date type     level
1    6820 2014-06-12   ST firsttime
2    6820 2015-06-11   ST       new
4   17413 2014-05-01   MC    active
5   38553 2014-06-12   LC    active
6   38553 2015-06-11   LC    active
8   52760 2014-10-24   YA       new
9  717841 2014-05-01   YA    active
11 717841 2014-12-02   MC    active
12 747187 2014-03-01   LC       new
14 747187 2014-05-12   MC    active

【讨论】：