R中的匹配合并

【问题标题】：Match-Merge in RR中的匹配合并
【发布时间】：2016-08-05 00:59:37
【问题描述】：

我正在努力将一些 SAS 程序迁移到 R 中，我很难找出完成这个看似简单的逻辑的最佳方法。

proc sort data=df1; by cridsessid mode refresh_key; run;
proc sort data=df2; by cridsessid mode refresh_key; run;

data df3;
  merge df1(in=xx) df2(in=yy);
  by cridsessid mode refresh_key;
  if xx and yy then do;
    cridsessid=catx(':',cridsessid,refresh_key2);
  end;
run;;

数据集很困难，因为在数百万行中可能只有几千行实际匹配xx 和yy，但基本输出如下所示：

df1

                           transId_app mode                 sm_bdt                 sm_edt
1 c3bca1af-ed0d-4403-9552-29758055f7a3 None 21MAR2016:07:07:56.611 21MAR2016:07:07:56.627
2 68f85148-6b75-49dc-90f9-5bb66b6a750b None 21MAR2016:07:07:56.940 21MAR2016:07:07:56.940
3 68f85148-6b75-49dc-90f9-5bb66b6a750b None 21MAR2016:07:07:56.940 21MAR2016:07:07:56.955
4 68f85148-6b75-49dc-90f9-5bb66b6a750b None 21MAR2016:07:07:56.940 21MAR2016:07:07:56.940
5 68f85148-6b75-49dc-90f9-5bb66b6a750b None 21MAR2016:07:07:56.924 21MAR2016:07:07:56.940
6 7a0c53a3-00b4-4b81-8238-24a738e5f4ed None 21MAR2016:07:08:33.003 21MAR2016:07:08:33.003
  mode_                                                                cridsessid
1 Pass1 0043dfb7-9a98-4b7c-9c04-5afc05580843:96530555-3568-468c-9bb4-ddd77278f1b2
2 Pass1 0043dfb7-9a98-4b7c-9c04-5afc05580843:96530555-3568-468c-9bb4-ddd77278f1b2
3 Pass1 0043dfb7-9a98-4b7c-9c04-5afc05580843:96530555-3568-468c-9bb4-ddd77278f1b2
4 Pass1 0043dfb7-9a98-4b7c-9c04-5afc05580843:96530555-3568-468c-9bb4-ddd77278f1b2
5 Pass1 0043dfb7-9a98-4b7c-9c04-5afc05580843:96530555-3568-468c-9bb4-ddd77278f1b2
6 Pass1 0043dfb7-9a98-4b7c-9c04-5afc05580843:96530555-3568-468c-9bb4-ddd77278f1b2
  refresh_key
1       7.125
2       7.125
3       7.125
4       7.125
5       7.125
6       7.142

带有dput() 的

structure(list(transId_app = c("c3bca1af-ed0d-4403-9552-29758055f7a3", 
"68f85148-6b75-49dc-90f9-5bb66b6a750b", "68f85148-6b75-49dc-90f9-5bb66b6a750b", 
"68f85148-6b75-49dc-90f9-5bb66b6a750b", "68f85148-6b75-49dc-90f9-5bb66b6a750b", 
"7a0c53a3-00b4-4b81-8238-24a738e5f4ed", "7a0c53a3-00b4-4b81-8238-24a738e5f4ed", 
"7a0c53a3-00b4-4b81-8238-24a738e5f4ed", "7a0c53a3-00b4-4b81-8238-24a738e5f4ed"
), mode = c("None", "None", "None", "None", "None", "None", "None", 
"None", "None"), sm_bdt = c("21MAR2016:07:07:56.611", "21MAR2016:07:07:56.940", 
"21MAR2016:07:07:56.940", "21MAR2016:07:07:56.940", "21MAR2016:07:07:56.924", 
"21MAR2016:07:08:33.003", "21MAR2016:07:08:33.003", "21MAR2016:07:08:32.988", 
"21MAR2016:07:08:32.957"), sm_edt = c("21MAR2016:07:07:56.627", 
"21MAR2016:07:07:56.940", "21MAR2016:07:07:56.955", "21MAR2016:07:07:56.940", 
"21MAR2016:07:07:56.940", "21MAR2016:07:08:33.003", "21MAR2016:07:08:33.003", 
"21MAR2016:07:08:33.003", "21MAR2016:07:08:33.003"), mode_ = c("Pass1", 
"Pass1", "Pass1", "Pass1", "Pass1", "Pass1", "Pass1", "Pass1", 
"Pass1"), cridsessid = c("0043dfb7-9a98-4b7c-9c04-5afc05580843:96530555-3568-468c-9bb4-ddd77278f1b2", 
"0043dfb7-9a98-4b7c-9c04-5afc05580843:96530555-3568-468c-9bb4-ddd77278f1b2", 
"0043dfb7-9a98-4b7c-9c04-5afc05580843:96530555-3568-468c-9bb4-ddd77278f1b2", 
"0043dfb7-9a98-4b7c-9c04-5afc05580843:96530555-3568-468c-9bb4-ddd77278f1b2", 
"0043dfb7-9a98-4b7c-9c04-5afc05580843:96530555-3568-468c-9bb4-ddd77278f1b2", 
"0043dfb7-9a98-4b7c-9c04-5afc05580843:96530555-3568-468c-9bb4-ddd77278f1b2", 
"0043dfb7-9a98-4b7c-9c04-5afc05580843:96530555-3568-468c-9bb4-ddd77278f1b2", 
"0043dfb7-9a98-4b7c-9c04-5afc05580843:96530555-3568-468c-9bb4-ddd77278f1b2", 
"0043dfb7-9a98-4b7c-9c04-5afc05580843:96530555-3568-468c-9bb4-ddd77278f1b2"
), refresh_key = c(7.125, 7.125, 7.125, 7.125, 7.125, 7.142, 
7.142, 7.142, 7.142)), .Names = c("transId_app", "mode", "sm_bdt", 
"sm_edt", "mode_", "cridsessid", "refresh_key"), class = "data.frame", row.names = c(NA, 
9L))

df2

              mode
1             None
2             None
3 LazyLoadUncached
4             None
5 LazyLoadUncached
6             None
                                                                 cridsessid refresh_key_
1 0043dfb7-9a98-4b7c-9c04-5afc05580843:96530555-3568-468c-9bb4-ddd77278f1b2      7.142+0
2 0043dfb7-9a98-4b7c-9c04-5afc05580843:96530555-3568-468c-9bb4-ddd77278f1b2      7.142+0
3 00e8a4c5-904d-46a4-a7b1-e4e6383bdd10:bf97d617-fbe8-4c4d-ab0e-c00d48bce120      0.317+0
4 00e8a4c5-904d-46a4-a7b1-e4e6383bdd10:bf97d617-fbe8-4c4d-ab0e-c00d48bce120      0.317+0
5 02063ca4-ccf3-4326-b87d-fe5ab13d2d7f:07783670-5ace-47bc-a707-db5b8064e241     5.658+17
6 02063ca4-ccf3-4326-b87d-fe5ab13d2d7f:07783670-5ace-47bc-a707-db5b8064e241     5.658+17
  refresh_key refresh_key2
1       7.142        7.142
2       7.142        7.142
3       0.317        0.317
4       0.317        0.317
5       5.658        5.658
6       5.658        5.658

带有dput()的

structure(list(mode = c("None", "None", "LazyLoadUncached", "None", 
"LazyLoadUncached", "None"), cridsessid = c("0043dfb7-9a98-4b7c-9c04-5afc05580843:96530555-3568-468c-9bb4-ddd77278f1b2", 
"0043dfb7-9a98-4b7c-9c04-5afc05580843:96530555-3568-468c-9bb4-ddd77278f1b2", 
"00e8a4c5-904d-46a4-a7b1-e4e6383bdd10:bf97d617-fbe8-4c4d-ab0e-c00d48bce120", 
"00e8a4c5-904d-46a4-a7b1-e4e6383bdd10:bf97d617-fbe8-4c4d-ab0e-c00d48bce120", 
"02063ca4-ccf3-4326-b87d-fe5ab13d2d7f:07783670-5ace-47bc-a707-db5b8064e241", 
"02063ca4-ccf3-4326-b87d-fe5ab13d2d7f:07783670-5ace-47bc-a707-db5b8064e241"
), refresh_key_ = c("7.142+0", "7.142+0", "0.317+0", "0.317+0", 
"5.658+17", "5.658+17"), refresh_key = c(7.142, 7.142, 0.317, 
0.317, 5.658, 5.658), refresh_key2 = c(7.142, 7.142, 0.317, 0.317, 
5.658, 5.658)), .Names = c("mode", "cridsessid", "refresh_key_", 
"refresh_key", "refresh_key2"), row.names = c(NA, 6L), class = "data.frame")

df3

                           transId_app mode                 sm_bdt                 sm_edt
1 7a0c53a3-00b4-4b81-8238-24a738e5f4ed None 21MAR2016:07:08:33.003 21MAR2016:07:08:33.003
2 7a0c53a3-00b4-4b81-8238-24a738e5f4ed None 21MAR2016:07:08:33.003 21MAR2016:07:08:33.003
3 7a0c53a3-00b4-4b81-8238-24a738e5f4ed None 21MAR2016:07:08:32.988 21MAR2016:07:08:33.003
4 7a0c53a3-00b4-4b81-8238-24a738e5f4ed None 21MAR2016:07:08:32.957 21MAR2016:07:08:33.003
  mode_                                                                      cridsessid
1 Pass1 0043dfb7-9a98-4b7c-9c04-5afc05580843:96530555-3568-468c-9bb4-ddd77278f1b2:7.142
2 Pass1 0043dfb7-9a98-4b7c-9c04-5afc05580843:96530555-3568-468c-9bb4-ddd77278f1b2:7.142
3 Pass1 0043dfb7-9a98-4b7c-9c04-5afc05580843:96530555-3568-468c-9bb4-ddd77278f1b2:7.142
4 Pass1 0043dfb7-9a98-4b7c-9c04-5afc05580843:96530555-3568-468c-9bb4-ddd77278f1b2:7.142
  refresh_key refresh_key_ refresh_key2
1       7.142      7.142+0        7.142
2       7.142      7.142+0        7.142
3       7.142      7.142+0        7.142
4       7.142      7.142+0        7.142

带有dput()的

structure(list(transId_app = c("7a0c53a3-00b4-4b81-8238-24a738e5f4ed", 
"7a0c53a3-00b4-4b81-8238-24a738e5f4ed", "7a0c53a3-00b4-4b81-8238-24a738e5f4ed", 
"7a0c53a3-00b4-4b81-8238-24a738e5f4ed"), mode = c("None", "None", 
"None", "None"), sm_bdt = c("21MAR2016:07:08:33.003", "21MAR2016:07:08:33.003", 
"21MAR2016:07:08:32.988", "21MAR2016:07:08:32.957"), sm_edt = c("21MAR2016:07:08:33.003", 
"21MAR2016:07:08:33.003", "21MAR2016:07:08:33.003", "21MAR2016:07:08:33.003"
), mode_ = c("Pass1", "Pass1", "Pass1", "Pass1"), cridsessid = c("0043dfb7-9a98-4b7c-9c04-5afc05580843:96530555-3568-468c-9bb4-ddd77278f1b2:7.142", 
"0043dfb7-9a98-4b7c-9c04-5afc05580843:96530555-3568-468c-9bb4-ddd77278f1b2:7.142", 
"0043dfb7-9a98-4b7c-9c04-5afc05580843:96530555-3568-468c-9bb4-ddd77278f1b2:7.142", 
"0043dfb7-9a98-4b7c-9c04-5afc05580843:96530555-3568-468c-9bb4-ddd77278f1b2:7.142"
), refresh_key = c(7.142, 7.142, 7.142, 7.142), refresh_key_ = c("7.142+0", 
"7.142+0", "7.142+0", "7.142+0"), refresh_key2 = c(7.142, 7.142, 
7.142, 7.142)), .Names = c("transId_app", "mode", "sm_bdt", "sm_edt", 
"mode_", "cridsessid", "refresh_key", "refresh_key_", "refresh_key2"
), row.names = c(NA, 4L), class = "data.frame")

我知道 dplyr 中的 semi_join()，并认为它可以有效地替换 if xx and yy，但 semi_join() 返回一个 data.frame，我不确定如何获得可以使用的逻辑向量做这样的事情：

df <- full_join(df1, df2, by="cridsessid", "mode", "refresh_key")
x <- {logical vector}
df[x, "cridsessid"] <- df[x, paste("cridsessid", "mode", sep=":")]

【问题讨论】：

  • 请提供minimal reproducible example 和预期输出
  • 该示例与 cressid 不匹配，并且还应该使用 dput 生成以允许粘贴到正在运行的 R 会话中。
  • 添加了 dput() 输出。谢谢你。就像我说的，..还在学习:)

标签： r dataframe merge sas dplyr


【解决方案1】：

在缺少少量输入集和输出数据集的情况下，SAS 将返回什么有点不清楚。 （SAS 合并返回什么样的连接？）我会想象 xx 和 yy 对于内部连接都是 TRUE。然后，您将刚刚将粘贴的 var1,":",var2 的值分配给 var1。我没有使用inner_join，因为这是基本函数merge的默认操作：

df3 <- merge(df1, df2, by="var1", "var2", "var3")

df3["var1"] <- with(df3, paste0(var1, var4, sep=":"))

【讨论】：

  • 很好，我错过了，并更新为full_join()，因为它应该是这样的。我还添加了一些小的示例数据。谢谢。
  • 如果您想要一个完全联接，则使用all=TRUE 作为merge 的参数，然后测试 NA 值，但您的示例至少有一个匹配标准没有匹配项。所以它不能真正用于有用的演示。
  • 测试 NA 值是我一直在寻找的关键。如此简单，我不敢相信我之前没有看到。非常感谢您的帮助。
猜你喜欢
  • 1970-01-01
  • 2015-11-22
  • 2018-02-04
  • 1970-01-01
  • 2023-03-31
  • 1970-01-01
  • 2012-05-23
  • 1970-01-01
  • 2015-11-22
相关资源
最近更新 更多
热门标签
Java Python linux javascript Mysql C# Docker 算法 前端 SpringBoot Redis Vue spring 设计模式 .net core .net kubernetes c++ 数据库 数据结构 大数据 js 机器学习 微服务 Android Go 程序员 面试 JVM ASP.net core 云原生 人工智能 后端 PHP git CSS golang k8s Nginx Django mybatis 深度学习 多线程 React 架构 devops 爬虫 云计算 Spring Boot LeetCode