無法通過data.table中的引用指定列as.Date

Question

在使用by =時，我正在分配一個新列作爲Date或IDate。這是創建一個整數列，而不是預期的Date。

require(data.table) 
dt <- data.table(date = as.IDate(sample(10000:11000, 10), 
           origin = "1970-01-01")) 
dt[, group := rep(1:2, 5)] 
print(dt) 

#   date group 
# 1: 1997-06-12  1 
# 2: 1998-02-19  2 
# 3: 1998-04-25  1 
# 4: 1998-01-27  2 
# 5: 1997-10-29  1 
# 6: 1998-05-08  2 
# 7: 1999-05-09  1 
# 8: 1999-06-26  2 
# 9: 1997-11-01  1 
# 10: 1997-07-19  2 

這工作：

dt[, min.date := min(date)] 
print(dt) 

#   date group min.date 
# 1: 1997-06-12  1 1997-06-12 
# 2: 1998-02-19  2 1997-06-12 
# 3: 1998-04-25  1 1997-06-12 
# 4: 1998-01-27  2 1997-06-12 
# 5: 1997-10-29  1 1997-06-12 
# 6: 1998-05-08  2 1997-06-12 
# 7: 1999-05-09  1 1997-06-12 
# 8: 1999-06-26  2 1997-06-12 
# 9: 1997-11-01  1 1997-06-12 
# 10: 1997-07-19  2 1997-06-12 

但這裏有一個問題：

dt[, min.group.date := as.IDate(min(date)), by = group] 
print(dt) 

#   date group min.date min.group.date 
# 1: 1997-06-12  1 1997-06-12   10024 
# 2: 1998-02-19  2 1997-06-12   10061 
# 3: 1998-04-25  1 1997-06-12   10024 
# 4: 1998-01-27  2 1997-06-12   10061 
# 5: 1997-10-29  1 1997-06-12   10024 
# 6: 1998-05-08  2 1997-06-12   10061 
# 7: 1999-05-09  1 1997-06-12   10024 
# 8: 1999-06-26  2 1997-06-12   10061 
# 9: 1997-11-01  1 1997-06-12   10024 
# 10: 1997-07-19  2 1997-06-12   10061 

min.group.date是數字，而不是Date。

dt[, class(min.group.date)] 

# [1] "numeric" 

如果我初始化列作爲Date或IDate，它按預期工作：

dt <- data.table(date = as.IDate(sample(10000:11000, 10), origin = "1970-01-01")) 
dt[, group := rep(1:2, 5)] 

dt[, min.group.date := as.IDate(NA)] 
dt[, min.group.date := min(date), by = group] 

dt[, class(min.group.date)] 
# [1] "IDate" "Date" 

Answer 1

保羅，如果你想要的是由最小日期組，這條線將做到這一點：

dt[,min(date),by=group] 

你應該看到（如下明顯的日期從你在你的例子不同，因爲「樣本」命令）：

group   V1 
1:  1 1997-11-19 
2:  2 1997-12-04 

如果你想看到的每一行，你可以加入表：

setkey(dt,group) #always good practice 
dt_min=dt[,min(date),by=group] 
setnames(dt_min,"V1","min.group.Date") #you should NOT use colnames (see help('setnames') 
dt[dt_min] 


    group  date min.group.Date 
1:  1 1999-01-30  1997-11-19 
2:  1 1999-11-27  1997-11-19 
3:  1 1999-11-11  1997-11-19 
4:  1 1997-11-19  1997-11-19 
5:  1 1999-05-06  1997-11-19 
6:  2 1999-07-11  1997-12-04 
7:  2 1997-12-04  1997-12-04 
8:  2 1998-07-28  1997-12-04 
9:  2 1998-10-23  1997-12-04 
10:  2 1998-06-01  1997-12-04