我一直在努力寻找一种相对简单的方法来使用R按组对日期范围内的事件进行计数。我认为，必须有一种比我尝试的方法更简单的方法。
我有6,000多个小组，每个小组中的ID范围从1到100，每个ID的开始日期和结束日期从1990年1月1日到今天。我想制作一个数据框，每列一组，每行一天，计算从2013年4月1日到2018年3月31日每天活动的ID数。出于明显的原因，在excel中使用countif不会削减它。
我试图以this question作为起点，例如：
df1 <- data.frame(group = c(1,1,2,3,3),
id = c(1,2,1,1,2),
startdate = c("2016-01-01","2016-04-04","2016-03-02","2016-08-01","2016-04-01"),
enddate = c("2016-04-04","2999-01-01","2016-05-02","2016-08-05","2999-01-01"))
report <- data.frame(date = seq(from = as.Date("2016-04-01"),by="1 day", length.out = 7))
report <- cbind(report,matrix(data=NA,nrow=7,ncol=3))
names(report) <- c('date',as.vector(unique(df1$group)))
daily <- function(i,...){
report[,i+1] <- sapply(report$date, function(x)
sum(as.Date(df1$startdate) < as.Date(x) &
as.Date(df1$enddate) > as.Date(x) &
df1$group == unique(df1$group)[i]))
}
for (i in unique(df1$group))
daily(i)
但是，这似乎什么也没做（也不抛出错误）。有没有更简单的方法可以做到这一点？我离基地远吗？此非程序员非常感谢您的帮助！
需要其他帮助：我正在尝试在下面的答案中修改Jaap的代码，以包括组开始时间和组结束时间，以便在组不活动时数据表显示NA。
示例数据：
df2 <- data.frame(group = c(1,1,2,3,3),
groupopendate = c("2016-04-02","2016-04-02","2016-04-01","2016-04-02","2016-04-02"),
groupclosedate = c("2016-04-08","2016-04-08","2016-04-10","2016-04-09","2016-04-09"),
id = c(1,2,1,1,2),
startdate = c("2016-04-02","2016-04-04","2016-04-03","2016-04-02","2016-04-05"),
enddate = c("2016-04-04","2016-04-06","2016-04-10","2016-04-08","2016-04-08"))
Jaap的解决方案给了我这个：
active grp1 grp2 grp3
1: 2016-04-02 1 0 1
2: 2016-04-03 1 1 1
3: 2016-04-04 1 1 1
4: 2016-04-05 1 1 2
5: 2016-04-06 0 1 2
6: 2016-04-07 0 1 2
但是，我想要的是这样的
active grp1 grp2 grp3
1: 2016-04-01 NA 0 NA
2: 2016-04-02 1 0 1
3: 2016-04-03 1 1 1
4: 2016-04-04 1 1 1
5: 2016-04-05 1 1 1
6: 2016-04-06 1 1 2
7: 2016-04-07 0 1 2
8: 2016-04-08 NA 1 0
9: 2016-04-09 NA 1 NA
10: 2016-04-10 NA NA NA
感谢您的帮助！
答案 0 :(得分：3)
使用data.table的可能替代解决方案：
# load the package & convert 'df1' to a data.table
library(data.table)
setDT(df1)
# convert the date columns to a date format
# not needed if they are
df1[, `:=` (startdate = as.Date(startdate), enddate = as.Date(enddate))]
# create a new data.table with the 'active' days
DT <- data.table(active = seq(from = as.Date("2016-04-01"), by = "day", length.out = 7))
# use a join and dcast to get the desired result
DT[df1
, on = .(active > startdate, active < enddate)
, allow = TRUE
, nomatch = 0
, .(active = x.active, group, id)
][, dcast(.SD, active ~ paste0("grp",group), value.var = "id", fun = length)]
给出：
active grp1 grp2 grp3 1: 2016-04-01 1 1 0 2: 2016-04-02 1 1 1 3: 2016-04-03 1 1 1 4: 2016-04-04 0 1 1 5: 2016-04-05 1 1 1 6: 2016-04-06 1 1 1 7: 2016-04-07 1 1 1
注意：我在paste0("grp",group)步骤中使用了group而不是dcast，因为它会导致更好的列名（最好不要仅使用数字值作为列名）< / p>
关于您的其他示例，您可以按照以下方法解决该问题：
setDT(df2)
df2[, c(2:3,5:6) := lapply(.SD, as.Date), .SDcols = c(2:3,5:6)]
DT <- data.table(active = seq(from = min(df2$groupopendate),
to = max(df2$groupclosedate),
by = "day"))
df2new <- df2[, .(active = seq.Date(startdate, enddate, by = "day"))
, by = .(group, id)
][, .N, by = .(group, active)
][df2[, .(active = seq.Date(groupopendate[1], groupclosedate[.N] - 1, by = "day"))
, by = .(group)]
, on = .(group, active)
][is.na(N), N := 0
][, dcast(.SD, active ~ paste0("grp",group))]
nms <- setdiff(names(df2new), "active")
DT[df2new
, on = .(active)
, (nms) := mget(paste0("i.",nms))][]
给出：
> DT active grp1 grp2 grp3 1: 2016-04-01 NA 0 NA 2: 2016-04-02 1 0 1 3: 2016-04-03 1 1 1 4: 2016-04-04 2 1 1 5: 2016-04-05 1 1 2 6: 2016-04-06 1 1 2 7: 2016-04-07 0 1 2 8: 2016-04-08 NA 1 2 9: 2016-04-09 NA 1 NA 10: 2016-04-10 NA 1 NA
答案 1 :(得分：1)
我知道了！像往常一样，一旦您发布问题，便会找到答案。当我可以将sapply放入for循环时，我通过放入函数使它过于复杂。
如果有人感兴趣：
for (i in unique(df1$group))
{report[,i+1] <-
sapply(report$date, function(x)
sum(as.Date(df1$startdate) < as.Date(x) &
as.Date(df1$enddate) > as.Date(x) &
df1$group == unique(df1$group)[i]))}