dd <- read.table(header = TRUE, text = "Surgeon Length 'Surg. Date'
John 75 2015-07-06
Max 120 2015-06-22
Max 190 2015-01-26
David 40 2015-11-04
David 25 2015-04-21
David 50 2015-12-11
Andrey 210 2015-03-15
Vincent 180 2015-01-30
Vincent 180 2015-06-10", check.names = FALSE)
我們可以設置Surgeon
的等級,使得r負責爲我們排序。如果我們列表外科醫生並對錶格進行排序,您可以看到表格名稱按照您的要求排序,因此我們只需將此順序設置爲levels(Surgeon)
的順序,而不是缺省(按字母順序)。
然後我們簡單地添加按日期排序的附加級別。
sort(tbl <- table(dd$Surgeon))
# Andrey John Max Vincent David
# 1 1 2 2 3
對於那些有關係,我們也可以與第一日期
(lvls <- names(tbl)[order(tbl, tapply(as.Date(dd$`Surg. Date`), dd$Surgeon, min))])
# [1] "Andrey" "John" "Max" "Vincent" "David"
dd$Surgeon <- factor(dd$Surgeon, levels = lvls)
dd[order(dd$Surgeon, dd$`Surg. Date`), ]
# Surgeon Length Surg. Date
# 7 Andrey 210 2015-03-15
# 1 John 75 2015-07-06
# 3 Max 190 2015-01-26
# 2 Max 120 2015-06-22
# 8 Vincent 180 2015-01-30
# 9 Vincent 180 2015-06-10
# 5 David 25 2015-04-21
# 4 David 40 2015-11-04
# 6 David 50 2015-12-11
與@ akrun的dplyr
解決方案添加一個排序的因素,你可以做一個類似的方法有更高的效率。
library('dplyr')
dd %>%
group_by(Surgeon) %>%
mutate(n=n()) %>%
ungroup() %>%
arrange(n, Surgeon, `Surg. Date`) %>%
select(-n)
# Surgeon Length Surg. Date
# (fctr) (int) (fctr)
# 1 Andrey 210 2015-03-15
# 2 John 75 2015-07-06
# 3 Max 190 2015-01-26
# 4 Max 120 2015-06-22
# 5 Vincent 180 2015-01-30
# 6 Vincent 180 2015-06-10
# 7 David 25 2015-04-21
# 8 David 40 2015-11-04
# 9 David 50 2015-12-11
或者,如果您訂購的因子水平上面一樣,你可以做
dd %>% arrange(Surgeon, `Surg. Date`)
與data.table
,你仍然可以使用表/因子水平的方法和設置按鍵,但我不知道這是data.table方式(即,唯一的開銷是這似乎是相當快的大載體的table
)
library('data.table')
dd$Surgeon <- factor(dd$Surgeon, levels = names(sort(table(dd$Surgeon))))
setDT(dd, key = c('Surgeon', 'Surg. Date'))
# Surgeon Length Surg. Date
# 1: Andrey 210 2015-03-15
# 2: John 75 2015-07-06
# 3: Max 190 2015-01-26
# 4: Max 120 2015-06-22
# 5: Vincent 180 2015-01-30
# 6: Vincent 180 2015-06-10
# 7: David 25 2015-04-21
# 8: David 40 2015-11-04
# 9: David 50 2015-12-11
不可再生;( – jangorecki