Question

有效过滤最后/最近一周的最佳方式是什么（基于它可能不是一整周的数据）。

library(lubridate)
library(dplyr)

df <- data.frame(dates =
c("2014-12-17","2014-12-18","2014-12-21","2014-12-25","2014-12-26",
  "2015-05-17","2015-05-18","2015-05-21","2015-05-25","2015-05-26",
  "2016-06-17","2016-06-18","2016-06-21","2016-06-25","2016-06-26"))


df <- df %>% mutate(dates = ymd(dates),
                    the.year = year(dates),
                    the.week = week(dates))

#Filter the last week (as may not be complete)

我可以提出像这样的解决方案

max.week <- df %>% filter(the.year == max(the.year)) %>%
  filter(the.week == max(the.week)) %>%
  group_by(the.year, the.week) %>%
  summarise(count= n()) %>%
  ungroup() %>%
  mutate(max.week = paste(the.year, the.week,sep="-")) %>% 
  select(max.week) %>%
  unlist(use.names = F)

df %>% filter(!paste(the.year, the.week, sep = "-") == max.week) 
   %>%

但必须有更简单的解决方案？

Answer 1

我能想到的最短的dplyr方式是

filter(df, !{yw <- interaction(the.year, the.week)} %in% yw[which.max(dates)])

但你可能想要将其分解以获得更好的易读性：

df %>% 
  mutate(yearweek = paste(the.year, the.week, sep = "-")) %>% 
  filter(!yearweek %in% yearweek[which.max(dates)])

删除!以达到相反的效果。

Answer 2

group_indices也可以提供帮助：

df %>% 
  filter(group_indices(., the.year, the.week) < max(group_indices(., the.year, the.week)))

也可以写成：

df %>% filter({id <- group_indices(., the.year, the.week)} < max(id))

或

df %>% 
  mutate(id = group_indices(., the.year, the.week)) %>% 
  filter(id < max(id))

Answer 3

试试这个：

df %>% transform(yw= the.year *100 + the.week) %>% filter(yw != max(yw)) %>% select(-yw)

或者，如果您的数据按日期排序似乎是这样的：

df %>% filter(the.year !=last(the.year) | the.week !=last(the.week))

Answer 4

使用dplyr，

的另一种可能性

df %>% 
arrange(dates) %>% 
filter(the.week != last(the.week) | the.year != last(the.year))

dplyr以编程方式过滤最近一周

4 个答案: