使用 R 中的多列将数据集拆分为两个数据框

Question

假设我的数据集如下所示：

working_data <- dplyr::data_frame("Date" = c("2015-01-01", "2015-01-01", "2015-01-02", "2015-01-03", "2015-01-04", "2015-01-04", "2015-01-04"),
                                  "Time" = c("15:01", "15:01", "21:04", "13:19", "07:15", "07:15", "07:15"),
                                  "SeizureTime" = c("0:10", "0:07", "0:11", "0:04", "0:08", "0:06", "0:07"),
                                  "ET" = c("0:35", "0:35", "0:04", "1:10", "3:35", "3:35", "3:35"),
                                  "ONumber" = c("(123)555-1234", "(123)555-1234", "(123)555-9999", "(000)555-9876", "(123)555-1111", "(123)555-1111", "(123)555-1111"),
                                  "TNumber" = c("(123)555-1234", "(123)555-1234", "(123)555-9999", "(000)555-9876", "(123)555-1111", "(123)555-1111", "(123)555-1111"),
                                  "CT" = c("a", "a", "b", "a", "b", "b", "b"))

我想从这些数据中提取可能重复的行。我这样做的方法如下：

while (nrow(working_data) != 0) {
          target_call <- working_data[1, ]
          working_data <- working_data[-1, ]
          similar_calls <- working_data %>% dplyr::filter(Date == target_call$Date,
                                                   Time == target_call$Time,
                                                   ET == target_call$ET,
                                                   ONumber == target_call$ONumber,
                                                   TNumber == target_call$TNumber)

第一个循环将设置target_call 等于working_data 的第一行，并将设置similar_calls 等于第二行。假设一切顺利......我遇到的问题是，一旦我在target_call 和similar_calls 上运行我的函数，我就不想再看到它们了。所以我想从working_data 中删除被拉入similar_calls 的数据。

填充target_call 和similar_calls 后，我需要确定哪些调用（如果有）与target_call 相同，然后进一步确定哪个是正确的选择，一旦我'选择了正确的调用，将其添加到名为 resolved_calls 的新数据集。如果similar_calls 中还有剩余呼叫，那么我需要重复选择呼叫的分析并将其中一个呼叫添加到resolved_calls。

我能想到的最佳方法是将数据拆分为两个单独的数据帧。但是当我处理多个列时，我不知道该怎么做。我唯一的选择是一个非常丑陋的 ifelse 语句，例如：

working_data$Group <- ifelse(working_data$Date == target_call$Date & ... & working_data$TNumber == target_call$TNumber, 1, 0)
similar_calls <- working_data %>% dplyr::filter(Group == 1)
working_data <- working_data %>% dplyr::filter(Group == 0)

有没有更好的方法来做到这一点？

Answer 1

您还没有真正描述您想对每个组做什么，但让我们假设您只想获取每组相似调用中的第一个元素。然后像duplicated 函数这样的东西可以很好地工作：

working_data[with(working_data, !duplicated(paste(Date, Time, ET, ONumber, TNumber))),]
# Source: local data frame [4 x 7]
# 
#         Date  Time SeizureTime    ET       ONumber       TNumber    CT
#        (chr) (chr)       (chr) (chr)         (chr)         (chr) (chr)
# 1 2015-01-01 15:01        0:10  0:35 (123)555-1234 (123)555-1234     a
# 2 2015-01-02 21:04        0:11  0:04 (123)555-9999 (123)555-9999     b
# 3 2015-01-03 13:19        0:04  1:10 (000)555-9876 (000)555-9876     a
# 4 2015-01-04 07:15        0:08  3:35 (123)555-1111 (123)555-1111     b

在 dplyr 语法中，您可以使用 group_by 按适当的元素进行分组，然后您可以使用 filter 和 row_number 来获取每个组中的第一个实例：

working_data %>%
  group_by(Date, Time, ET, ONumber, TNumber) %>%
  filter(row_number() == 1)
# Source: local data frame [4 x 7]
# Groups: Date, Time, ET, ONumber, TNumber [4]
# 
#         Date  Time SeizureTime    ET       ONumber       TNumber    CT
#        (chr) (chr)       (chr) (chr)         (chr)         (chr) (chr)
# 1 2015-01-01 15:01        0:10  0:35 (123)555-1234 (123)555-1234     a
# 2 2015-01-02 21:04        0:11  0:04 (123)555-9999 (123)555-9999     b
# 3 2015-01-03 13:19        0:04  1:10 (000)555-9876 (000)555-9876     a
# 4 2015-01-04 07:15        0:08  3:35 (123)555-1111 (123)555-1111     b

如果您想更一般地处理组，您可以使用group_by 然后summarize 以不同方式总结组：

# Take text data in format mm:ss and return the number of seconds
secs <- function(x) {
  spl <- strsplit(x, ":")
  60*as.numeric(sapply(spl, "[", 1)) + as.numeric(sapply(spl, "[", 2))
}
working_data %>%
  group_by(Date, Time, ET, ONumber, TNumber) %>% 
  summarize(meanSeizure=mean(secs(SeizureTime)))
# Source: local data frame [4 x 6]
# Groups: Date, Time, ET, ONumber [?]
# 
#         Date  Time    ET       ONumber       TNumber meanSeizure
#        (chr) (chr) (chr)         (chr)         (chr)       (dbl)
# 1 2015-01-01 15:01  0:35 (123)555-1234 (123)555-1234         8.5
# 2 2015-01-02 21:04  0:04 (123)555-9999 (123)555-9999        11.0
# 3 2015-01-03 13:19  1:10 (000)555-9876 (000)555-9876         4.0
# 4 2015-01-04 07:15  3:35 (123)555-1111 (123)555-1111         7.0