R中的组重叠(市场份额/相关性)

Question

我在R中有用户级别的数据，它显示了用户与之交互的不同组。我想要弄清楚的是这些群体之间的重叠。我有一个下面的数据样本：

Group UserID
A     User1
B     User1
D     User1
A     User2

正如你所看到的，User1已经与3个群组进行了交互，而User2只与A组进行了交互。我想知道的是每个群组所拥有的用户的“市场份额”。例如，A组有100,000个用户只与A组交互，10,000个用户与A&B交互，5,000个用户与A，B和C交互，等等。显然有很多组合。

有没有办法用tidyr/dplyr来计算呢？大约有100万个用户和600个组。每个用户都有可能与某些组进行交互，但不是所有组。

谢谢!

Answer 1

如果你只想知道其他交互的数量，这是可行的：

df <- data.frame(Group = c("A", "B", "D", "A", "C", "D", "D", "B", "C"),
                 UserID = c("User1", "User1", "User1", "User2", "User2", "User3", "User4", "User5", "User5"))


library(tidyverse)

df %>% 
  group_by(Group, UserID) %>% # make sure there are no double entries
  summarise() %>%             # make sure there are no double entries
  group_by(UserID) %>%
  mutate(NGroups = n()) %>%   # how many interactions has this user
  ungroup() %>% 
  group_by(Group, NGroups) %>% 
  summarise(N = n()) %>%      # count for each Group - NInteractions combination the frequency
  ungroup() %>% 
  pivot_wider(names_from = NGroups, values_from = N)

如果您想要每个单独的组组合的计数，您应该可以开始使用;)：

df %>% 
  group_by(Group, UserID) %>% 
  summarise() %>% 
  group_by(UserID) %>%
  mutate(GroupsString = paste0(Group, collapse="")) %>% 
  ungroup() %>% 
  group_by(Group, GroupsString) %>% 
  summarise(N = n()) %>% 
  ungroup()