CODEWITHSUNDEEP
rdataframemultiple-conditions
RasK
RasK

Reputation: 11

How to create a new column in dataframe A, based on multiple conditions in dataframe B

I guess I don´t see the forest through the trees... Thus, I would like to ask for help.

Data: for dput() output see at the end of the question.

-Dataframe dfA with the columns: ID; ts. ts being POSIXct

-Dataframe dfB with the columns: ID; start; end; state_id. ID (corresponding to the ID in dfA), start (POSIXct), end (POSIXct), state_id.

The task:

I would like to create a new column in dfA with values 1/0 based on a condition. The condition in words: If the ID in dfA and dfB matches, and if the timestamp dfA$ts is in between or equal to dfB$start and dfB$end, then the value 1 should be written into dfA$x, otherwise a 0 should be there.

I think the code should look somewhat like this:

dfA$x <- iflese( dfA$ID == dfB$ID & dfA$ts >= dfB$start & dfA$ts <= dfB$end, 1, 0)

Thanks for the help already in advance.

dput(dfB):

structure(list(ID = c(1151L, 1151L, 1150L, 1150L, 1150L, 1150L, 1152L, 1152L, 1152L, 1345L), start = structure(c(1443142500, 1443144600, 1442934900, 1442942400, 1442944800, 1442946300, 1443103500, 1443132600, 1443137400, 1443389400), class = c("POSIXct", "POSIXt" )), end = structure(c(1443143400, 1443145500, 1442935500, 1442943000, 1442945400, 1442950200, 1443106200, 1443134100, 1443140100, 1443392400 ), class = c("POSIXct", "POSIXt")), state_id = c(1L, 2L, 1L, 2L, 3L, 4L, 1L, 2L, 3L, 1L)), row.names = c(NA, -10L), class = "data.frame")

dput(dfA):

structure(list(ID = c(1151L, 1151L, 1151L, 1151L, 1151L, 1151L, 1151L, 1151L, 1151L, 1151L, 1151L, 1151L, 1151L, 1151L, 1151L, 1151L, 1151L, 1151L, 1151L, 1151L, 1150L, 1150L, 1150L, 1150L, 1150L, 1150L, 1150L, 1150L, 1150L, 1150L, 1150L, 1150L, 1152L, 1152L, 1152L, 1152L, 1152L, 1152L, 1152L, 1152L, 1152L, 1152L, 1152L, 1152L, 1152L, 1152L, 1152L, 1152L, 1345L, 1345L, 1345L, 1345L, 1345L, 1345L, 1345L, 1345L, 1345L, 1345L), ts = structure(c(1443141300, 1443141600, 1443141900, 1443142200, 1443142500, 1443142800, 1443143100, 1443143400, 1443143700, 1443144000, 1443144300, 1443144600, 1443144900, 1443145200, 1443145500, 1443145800, 1443146100, 1443146400, 1443146700, 1443147000, 1442934900, 1442935200, 1442935500, 1442935800, 1442936100, 1442936400, 1442936700, 1442937000, 1442937300, 1442937600, 1442937900, 1442938200, 1443103500, 1443103800, 1443104100, 1443104400, 1443104700, 1443105000, 1443105300, 1443105600, 1443105900, 1443106200, 1443106500, 1443106800, 1443107100, 1443107400, 1443107700, 1443108000, 1443369300, 1443369600, 1443369900, 1443370200, 1443370500, 1443370800, 1443371100, 1443371400, 1443371700, 1443372000), class = c("POSIXct", "POSIXt" ))), row.names = c(NA, -58L), class = "data.frame")

Upvotes: 0

Views: 70

Answers (3)

Edward
Edward

Reputation: 18493

This is an updating non-equi join:

library(data.table)

setDT(dfA); setDT(dfB)

dfA[dfB, match := 1, on=.(ID=ID, ts>=start, ts<=end)][, match:=ifelse(is.na(match), 0, match)]

      ID                  ts match
 1: 1151 2015-09-25 07:35:00     0
 2: 1151 2015-09-25 07:40:00     0
 3: 1151 2015-09-25 07:45:00     0
 4: 1151 2015-09-25 07:50:00     0
 5: 1151 2015-09-25 07:55:00     1
 6: 1151 2015-09-25 08:00:00     1
 7: 1151 2015-09-25 08:05:00     1
 8: 1151 2015-09-25 08:10:00     1
 9: 1151 2015-09-25 08:15:00     0
10: 1151 2015-09-25 08:20:00     0
...

Upvotes: 0

linog
linog

Reputation: 6226

You can use data.table, conditional replacement with :=:

data.table::setDT(dfA)
dfA[,value := 0L]
dfA[(get('ID') == dfB$ID) & (get('ts') >= dfB$start) & (get('ts') <= dfB$end), value := 1L]

Note that I put column names from dfA in a get call to avoid confusion with dfB columns.

In this situation dfA and dfB must have the same number of rows. If they don't, use a merge based on ID column

Upvotes: 1

Ronak Shah
Ronak Shah

Reputation: 388797

We can left_join dfA and dfB by 'ID', group_by each ID and ts and check if any value lies in the range for that group.

library(dplyr)

dfA %>%
  left_join(dfB, by = 'ID') %>%
  group_by(ID, ts) %>% 
  summarise(x = +any(ts >= start & ts <= end))


#      ID ts                      x
#   <int> <dttm>              <int>
# 1  1150 2015-09-22 23:15:00     1
# 2  1150 2015-09-22 23:20:00     1
# 3  1150 2015-09-22 23:25:00     1
# 4  1150 2015-09-22 23:30:00     0
# 5  1150 2015-09-22 23:35:00     0
# 6  1150 2015-09-22 23:40:00     0
# 7  1150 2015-09-22 23:45:00     0
# 8  1150 2015-09-22 23:50:00     0
# 9  1150 2015-09-22 23:55:00     0
#10  1150 2015-09-23 00:00:00     0
# … with 48 more rows

Upvotes: 0

Related Questions