R总开/关时间

我有一个数据集，带有时间戳和流量计的流量数据（加仑/分钟），每2分钟记录一次数据

df <- structure(list(dt = structure(c(1519891200, 1519891320, 1519891440, 1519891560, 
                1519891680, 1519891800, 1519891920, 1519892040, 1519892160, 1519892280,
                1519892400, 1519892520, 1519892640, 1519892760, 1519892880), 
       class = c("POSIXct", "POSIXt"), tzone = ""), 
       gpm = c(0, 0, 0, 50, 50, 50, 50, 50, 0, 0, 80, 80, 80, 0, 0)), 
       .Names = c("dt", "gpm"), 
       row.names = c(NA, 15L), 
       class = "data.frame")

#                     dt gpm
# 1  2018-03-01 03:00:00   0
# 2  2018-03-01 03:02:00   0
# 3  2018-03-01 03:04:00   0
# 4  2018-03-01 03:06:00  50
# 5  2018-03-01 03:08:00  50
# 6  2018-03-01 03:10:00  50
# 7  2018-03-01 03:12:00  50
# 8  2018-03-01 03:14:00  50
# 9  2018-03-01 03:16:00   0
# 10 2018-03-01 03:18:00   0
# 11 2018-03-01 03:20:00  80
# 12 2018-03-01 03:22:00  80
# 13 2018-03-01 03:24:00  80
# 14 2018-03-01 03:26:00   0
# 15 2018-03-01 03:28:00   0

---

在

dplyr

中使用

cumsum

：

library(dplyr)
df %>% mutate(id = cumsum((gpm + c(0, lag(gpm)[-1])) == 0)) %>% 
  filter(gpm != 0) %>% group_by(id) %>% 
  mutate_at(vars(dt), .funs = list(on = min, off = max)) %>% 
  summarise_all(list(~mean(., na.rm=TRUE))) %>%
  select(-dt,-id,-gpm,gpm)

或者在

数据中使用
rle
和
cumsum
。表

：

library(data.table)
setDT(df)[, id := {
  rid <-rle(+(df$gpm + c(0,lag(df$gpm)[-1]) != 0 ))
  rid$values <- cumsum(rid$values) * rid$values
  inverse.rle(rid) * +(df$gpm!=0)
        }][gpm !=0
          ][ , list(on = min(dt), off = max(dt), gpm = mean(gpm)) , 
               by = .(id)][,id:=NULL][]

library(data.table)
setDT(df)[, id := {
  rid <-rle(+(df$gpm + c(0,lag(df$gpm)[-1]) != 0 ))
  rid$values <- cumsum(rid$values) * rid$values
  inverse.rle(rid) * +(df$gpm!=0)
        }][gpm !=0
          ][ , list(on = min(dt), off = max(dt), gpm = mean(gpm)) , 
               by = .(id)][,id:=NULL][]

#                    on                 off gpm
# 1 2018-03-01 03:06:00 2018-03-01 03:14:00  50
# 2 2018-03-01 03:20:00 2018-03-01 03:24:00  80