CODEWITHSUNDEEP
rdataframesplitsubset
Oumab10
Oumab10

Reputation: 716

Split Data frame based on row index

I want to split my data frame by looping through rows and subsetting 

indices = (diff(Data$Time>1800))
for (i in 1:length(indices)){
if(indices[i]==TRUE){
##### I need a function to split data by row index
}}

I tried

lst <- split(Data, as.factor(diff(Data$Time>1800)==TRUE))

But it returns only rows that are true, but what i want is to split each time it faces true.

This is what I have

    Time        temp
7/1/17 13:45:34  56
7/1/17 13:45:37  68
7/1/17 13:45:39  98
7/1/17 13:45:40  99
7/1/17 13:45:46  97
7/1/17 14:16:29  48
7/1/17 14:16:30  78
7/1/17 14:16:31  66
7/1/17 14:17:34  93
7/1/17 14:17:39  98
7/1/17 14:17:40  98
7/1/17 14:17:44  93
7/1/17 14:47:10  54
7/1/17 14:47:12  67
7/1/17 14:47:16  69
7/1/17 14:47:18  95
7/1/17 14:47:19  95
7/1/17 14:47:23  96
7/1/17 14:47:28  96
7/1/17 14:47:30  99

This is what I want 

Time        temp
7/1/17 13:45:34  56
7/1/17 13:45:37  68
7/1/17 13:45:39  98
7/1/17 13:45:40  99
7/1/17 13:45:46  97

Time        temp
7/1/17 14:16:29  48
7/1/17 14:16:30  78
7/1/17 14:16:31  66
7/1/17 14:17:34  93
7/1/17 14:17:39  98
7/1/17 14:17:40  98
7/1/17 14:17:44  93

Time        temp
7/1/17 14:47:10  54
7/1/17 14:47:12  67
7/1/17 14:47:16  69
7/1/17 14:47:18  95
7/1/17 14:47:19  95
7/1/17 14:47:23  96
7/1/17 14:47:28  96
7/1/17 14:47:30  99

Is it possible to split my data by storing these indexes in a vector and then splitting data frame based on this vector, which means whenever the row number is equal to our value " i " split the data frame at that row. So as to have multiple data frames.

Upvotes: 2

Views: 2100

Answers (1)

akrun
akrun

Reputation: 887118

With the new dataset, it seems like instead of 1800, it should be 1700

library(dplyr)
library(purrr)
library(lubridate)
Data %>% 
    mutate(Time = dmy_hms(Time), 
           grp = cumsum(Time - lag(Time, default = Time[1L])> 1700)) %>%
    split(.$grp) %>% 
    map(~ .x %>%
            select(-grp))
#$`0`
#                 Time temp
#1 2017-01-07 13:45:34   56
#2 2017-01-07 13:45:37   68
#3 2017-01-07 13:45:39   98
#4 2017-01-07 13:45:40   99
#5 2017-01-07 13:45:46   97

#$`1`
#                  Time temp
#6  2017-01-07 14:16:29   48
#7  2017-01-07 14:16:30   78
#8  2017-01-07 14:16:31   66
#9  2017-01-07 14:17:34   93
#10 2017-01-07 14:17:39   98
#11 2017-01-07 14:17:40   98
#12 2017-01-07 14:17:44   93

#$`2`
#                  Time temp
#13 2017-01-07 14:47:10   54
#14 2017-01-07 14:47:12   67
#15 2017-01-07 14:47:16   69
#16 2017-01-07 14:47:18   95
#17 2017-01-07 14:47:19   95
#18 2017-01-07 14:47:23   96
#19 2017-01-07 14:47:28   96
#20 2017-01-07 14:47:30   99

Similar option with base R would be

split(Data, cumsum(c(0, diff(as.POSIXct(Data$Time,
             format = "%d/%m/%y %H:%M:%S", tz = 'GMT'))) > 1700))

Upvotes: 1

Related Questions