使用R进行数据重组

Question

我有一个如下所示的数据集(dat):

 Person       IPaddress
36598035    222.999.22.99
36598035    222.999.22.99
36598035    222.999.22.99
36598035    222.999.22.99
36598035    222.999.22.99
36598035    444.666.44.66
37811171    111.88.111.88
37811171    111.88.111.88
37811171    111.88.111.88
37811171    111.88.111.88
37811171    111.88.111.88

它反映了在一段时间内登录网站的个人的实例.我需要数据看起来像这样:

Person        IPaddress      Number of Logins
36598035    222.999.22.99           6
37811171    111.88.111.88           5

因此,对于同一个人而言,不是多个条目,每个人只有一行,并且计算他们登录的次数.

此外,您将在我的示例中注意到,人员36598035在多于1个IP地址下登录.发生这种情况时,我希望最终数据集中的IP地址反映模式IP地址 - 换句话说,个人最常登录的IP地址.

Answer 1

这是一种方法.

library(dplyr)

mydf %>%
    group_by(Person, IPaddress) %>% # For each combination of person and IPaddress
    summarize(freq = n()) %>% # Get total number of log-in
    arrange(Person, desc(freq)) %>% # The most frequent IP address is in the 1st row for each user
    group_by(Person) %>% # For each user
    mutate(total = sum(freq)) %>% # Get total number of log-in
    select(-freq) %>% # Remove count
    do(head(.,1)) # Take the first row for each user

#    Person     IPaddress total
#1 36598035 222.999.22.99     6
#2 37811171 111.88.111.88     5

UPDATE

dplyr0.3现在出来了.所以,你也可以做以下事情.使用时只需缩短一行count.我也用slice@aosmith推荐.

mydf %>%
    count(Person, IPaddress) %>%
    arrange(Person, desc(n)) %>%
    group_by(Person) %>%
    mutate(total = sum(n)) %>%
    select(-n) %>%
    slice(1)