Hi,
I'm sorry that it takes me so much time to respond, finally yesterday I got
time to try your suggestions. Thank you for them!
I tried both, they give the same results, but in both there are some things
I still need to solve. I would appreciate your help.
I include a little bigger dataframe (test2, in the end of this email), with
more differencies in variables, to be able to better explain what I would
like to calculate in addition.
*Jim's code:*
I needed to make some changes in assigning the key. Yours worked ok for
that small "test" data, but when I tried it on my dataframe which has
around 25000rows, it didn't work properly.
test2$key[test2$act == 0] <- 1
test2$key[test2$act > 0 & test2$act < 200] <- 2
test2$key[test2$act == 200] <- 3
# this works ok
test2$resChange <- cumsum(c(1, abs(diff(test2$key))))
test2$res <- ave(test2$resChange, test2$resChange, FUN = length)
# I added new column by jul date
test2$resJ <- ave(test2$resChange, test2$resChange, test2$juln, FUN =
length)
# this works fine as well, for dividing between day 0 and day 1
test2$resJD <- ave(test2$resChange, test2$resChange, test2$juln, test2$day,
FUN = length)
# resume
test2Resume <- test2[ , list(maxres = max(res)
, minres = min(res)
, sumres = length(unique(resChange)))
, keyby = c('day', 'key')]
# change 'key'
test2Resume_day$key <- c('0', '1-199', '200')[test2Resume_day$key]
test2Resume_day
day key maxres minres sumres
1: 0 0 2 2 3
2: 0 1-199 3 1 9
3: 0 200 6 1 7
4: 1 0 1 1 1
5: 1 1-199 10 1 7
6: 1 200 6 1 6
# resume by juln
test2Resume_jul <- test2[ , list(maxres = max(res)
, minres = min(res)
, sumres = length(unique(resChange)))
, keyby = c('juln', 'key')] # by juln
# change 'key'
test2Resume_jul$key <- c('0', '1-199', '200')[test2Resume_jul$key]
test2Resume_jul
juln key maxres minres sumres
1: 15173 0 2 2 1
2: 15173 1-199 3 1 7
3: 15173 200 6 1 6
4: 15174 0 2 1 3
5: 15174 1-199 10 1 8
6: 15174 200 6 1 6
It is ok, but what I would like to get is resume for juln and for variable
day (0 and 1) aswell.
Like this:
juln day key maxres minres sumres
15173 0 0
15173 0 1-199
15173 0 200
15173 1 0
15173 1 1-199
15173 1 200
15174 0 0
15174 0 1-199
15174 0 200
15174 1 0
15174 1 1-199
15174 1 200
...
The other thing is that the "sumres" I would like to calculate like a sum
of values of occurencies for each "key".
For example, if in the test2 dataframe res values for key 200 (juln 15173)
are 1, 1, 2,2,1,2 the sumres should be 9 (1+1+2+2+1+2), not 6 (which I
suppose come form sum of number of unique occurencies).
*Petr's code:*
This works fine also, the thing is that doing the aggregation I would need
the intervals to be like this
[0, 1)
[1, 199]
(199, 200]
what I don't know if is possible... I checked the hepl for cut, but I found
that it can be closed just right or left...
Thank you very much for your time and sharing your knowledge!
Zuzana
## here is the bigger test2 dataframe
> dput(test2)
structure(list(daten = structure(c(15173, 15173, 15173, 15173,
15173, 15173, 15173, 15173, 15173, 15173, 15173, 15173, 15173,
15173, 15173, 15173, 15173, 15173, 15173, 15173, 15173, 15173,
15174, 15174, 15174, 15174, 15174, 15174, 15174, 15174, 15174,
15174, 15174, 15174, 15174, 15174, 15174, 15174, 15174, 15174,
15174, 15174, 15174, 15174, 15174, 15174, 15174, 15174, 15174,
15174, 15174, 15174, 15174, 15174, 15174, 15174, 15174, 15174,
15174, 15174), class = "Date"), juln = c(15173, 15173, 15173,
15173, 15173, 15173, 15173, 15173, 15173, 15173, 15173, 15173,
15173, 15173, 15173, 15173, 15173, 15173, 15173, 15173, 15173,
15173, 15174, 15174, 15174, 15174, 15174, 15174, 15174, 15174,
15174, 15174, 15174, 15174, 15174, 15174, 15174, 15174, 15174,
15174, 15174, 15174, 15174, 15174, 15174, 15174, 15174, 15174,
15174, 15174, 15174, 15174, 15174, 15174, 15174, 15174, 15174,
15174, 15174, 15174), fen = c("win", "win", "win", "win", "win",
"win", "win", "win", "win", "win", "win", "win", "win", "win",
"win", "win", "win", "win", "win", "win", "win", "win", "win",
"win", "win", "win", "win", "win", "win", "win", "win", "win",
"win", "win", "win", "win", "win", "win", "win", "win", "win",
"win", "win", "win", "win", "win", "win", "win", "win", "win",
"win", "win", "win", "win", "win", "win", "win", "win", "win",
"win"), night = structure(c(1310962792, 1310963392, 1310963992,
1310964592, 1310965192, 1310965792, 1310966392, 1310966992, 1310967592,
1310968192, 1310968792, 1310969392, 1310969992, 1310970592, 1310971192,
1310971792, 1310972392, 1310972992, 1310973592, 1310974192, 1310974792,
1310975392, 1311107991, 1311108591, 1311109191, 1311109791, 1311110391,
1311110991, 1311111591, 1311112191, 1311112791, 1311113391, 1311113991,
1311114591, 1311115191, 1311115791, 1311116391, 1311116991, 1311117591,
1311118191, 1311118791, 1311119391, 1311119991, 1311034191, 1311034791,
1311035391, 1311035991, 1311036591, 1311037191, 1311037791, 1311038391,
1311038991, 1311039591, 1311040191, 1311040791, 1311041391, 1311041991,
1311042591, 1311043191, 1311043791), class = c("POSIXct", "POSIXt"
), tzone = "GMT"), day = c(1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0), act = c(196, 200, 199, 200, 197, 198, 197,
200, 200, 197, 200, 200, 198, 200, 1, 1, 0, 0, 1, 2, 200, 200,
200, 200, 200, 200, 199, 61, 0, 194, 198, 198, 196, 193, 194,
193, 197, 198, 199, 200, 197, 199, 199, 200, 198, 200, 200, 198,
200, 34, 1, 1, 0, 0, 199, 200, 199, 7, 0, 0)), .Names = c("daten",
"juln", "fen", "night", "day", "act"), row.names = 9990:10049, class =
"data.frame")
On 29 April 2013 14:35, PIKAL Petr <petr.pi...@precheza.cz> wrote:
> Hi
>
> rrr<-rle(as.numeric(cut(test$act, c(0,1,199,200), include.lowest=T)))
> test$res <- rep(rrr$lengths, rrr$lengths)
>
> If you put it in function
>
> fff<- function(x, limits=c(0,1,199,200)) {
> rrr<-rle(as.numeric(cut(x, limits, include.lowest=T)))
> res <- rep(rrr$lengths, rrr$lengths)
> res
> }
>
> you can use split/lapply approach
>
> test$res2<-unlist(lapply(split(test$act, factor(test$day, levels=c(1,0))),
> fff))
>
> Beware of correct ordering of days in output. Without correct leveling of
> factor 0 precedes 1.
>
> And for the last part probably aggregate can be the way.
>
> > aggregate(test$res, list(test$jul, cut(test$act, c(0,1,199,200),
> include.lowest=T)), max)
> Group.1 Group.2 x
> 1 14655 [0,1] 4
> 2 14655 (1,199] 3
> 3 14655 (199,200] 3
> > aggregate(test$res, list(test$jul, cut(test$act, c(0,1,199,200),
> include.lowest=T)), min)
> Group.1 Group.2 x
> 1 14655 [0,1] 4
> 2 14655 (1,199] 1
> 3 14655 (199,200] 2
>
> Regards
> Petr
>
> > -----Original Message-----
> > From: r-help-boun...@r-project.org [mailto:r-help-bounces@r-
> > project.org] On Behalf Of zuzana zajkova
> > Sent: Monday, April 29, 2013 12:45 PM
> > To: r-help@r-project.org
> > Subject: [R] Counting number of consecutive occurrences per rows
> >
> > Hi,
> >
> > I would appreciate if somebody could help me with following
> > calculation.
> > I have a dataframe, by 10 minutes time, for mostly one year data. This
> > is small example:
> >
> > > dput(test)
> > structure(list(jul = structure(c(14655, 14655, 14655, 14655, 14655,
> > 14655, 14655, 14655, 14655, 14655, 14655, 14655, 14655, 14655, 14655,
> > 14655), origin = structure(0, class = "Date")),
> > time = structure(c(1266258354, 1266258954, 1266259554, 1266260154,
> > 1266260754, 1266261354, 1266261954, 1266262554, 1266263154,
> > 1266263754, 1266264354, 1266264954, 1266265554, 1266266154,
> > 1266266754, 1266267354), class = c("POSIXct", "POSIXt"), tzone =
> > "GMT"),
> > act = c(130, 23, 45, 200, 200, 200, 199, 150, 0, 0, 0, 0,
> > 34, 200, 200, 145), day = c(1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
> > 0, 0, 0, 0, 0, 0)), .Names = c("jul", "time", "act", "day"
> > ), class = "data.frame", row.names = c(510L, 512L, 514L, 516L, 518L,
> > 520L, 522L, 524L, 526L, 528L, 530L, 532L, 534L, 536L, 538L,
> > 540L))
> >
> > Looks like this:
> >
> > > test
> > jul time act day
> > 510 14655 2010-02-15 18:25:54 130 1
> > 512 14655 2010-02-15 18:35:54 23 1
> > 514 14655 2010-02-15 18:45:54 45 1
> > 516 14655 2010-02-15 18:55:54 200 1
> > 518 14655 2010-02-15 19:05:54 200 1
> > 520 14655 2010-02-15 19:15:54 200 1
> > 522 14655 2010-02-15 19:25:54 199 1
> > 524 14655 2010-02-15 19:35:54 150 1
> > 526 14655 2010-02-15 19:45:54 0 1
> > 528 14655 2010-02-15 19:55:54 0 1
> > 530 14655 2010-02-15 20:05:54 0 0
> > 532 14655 2010-02-15 20:15:54 0 0
> > 534 14655 2010-02-15 20:25:54 34 0
> > 536 14655 2010-02-15 20:35:54 200 0
> > 538 14655 2010-02-15 20:45:54 200 0
> > 540 14655 2010-02-15 20:55:54 145 0
> >
> >
> > What I would like to calculate is the number of consecutive occurrences
> > of values 200, 0 and together values from 1 til 199 (in fact the
> > values that differ from 200 and 0) in column "act".
> >
> > I would like to get something like this (result$res)
> >
> > > result
> > jul time act day res res2
> > 510 14655 2010-02-15 18:25:54 130 1 3 3
> > 512 14655 2010-02-15 18:35:54 23 1 3 3
> > 514 14655 2010-02-15 18:45:54 45 1 3 3
> > 516 14655 2010-02-15 18:55:54 200 1 3 3
> > 518 14655 2010-02-15 19:05:54 200 1 3 3
> > 520 14655 2010-02-15 19:15:54 200 1 3 3
> > 522 14655 2010-02-15 19:25:54 199 1 2 2
> > 524 14655 2010-02-15 19:35:54 150 1 2 2
> > 526 14655 2010-02-15 19:45:54 0 1 4 2
> > 528 14655 2010-02-15 19:55:54 0 1 4 2
> > 530 14655 2010-02-15 20:05:54 0 0 4 2
> > 532 14655 2010-02-15 20:15:54 0 0 4 2
> > 534 14655 2010-02-15 20:25:54 34 0 1 1
> > 536 14655 2010-02-15 20:35:54 200 0 2 2
> > 538 14655 2010-02-15 20:45:54 200 0 2 2
> > 540 14655 2010-02-15 20:55:54 145 0 1 1
> >
> > And if possible, distinguish among day==1 and day==0 (see the "act"
> > values of 0 for example), results as in result$res2.
> >
> > After it I would like to make a resume table per days (jul):
> > where maxres is max(result$res) for the "act" value where minres is
> > min(result$res) for the "act" value where sumres is sum(result$res) for
> > the "act" value (for example, if the 200 value ocurrs in different
> > times per day(jul) consecutively 3, 5, 1, 6 and 7 times the sumres
> > would be 3+5+1+6+7= 22)
> >
> > something like this (this are made up numbers):
> >
> > jul act maxres minres sumres
> > 14655 0 4 1 25
> > 14655 200 3 2 48
> > 14655 1-199 3 1 71
> > 14656 0 8 2 38
> > 14656 200 15 3 60
> > 14656 1-199 11 4 46
> > ...
> > (theoretically the sum of sumres per day(jul) should be 144)
> >
> > ____________
> > > sessionInfo()
> > R version 2.15.2 (2012-10-26)
> > Platform: x86_64-apple-darwin9.8.0/x86_64 (64-bit) ____________
> >
> > I hope my explanation is sufficient. I appreciate any hint.
> > Thank you,
> >
> > Zuzana
> >
> > [[alternative HTML version deleted]]
> >
> > ______________________________________________
> > R-help@r-project.org mailing list
> > https://stat.ethz.ch/mailman/listinfo/r-help
> > PLEASE do read the posting guide http://www.R-project.org/posting-
> > guide.html
> > and provide commented, minimal, self-contained, reproducible code.
>
[[alternative HTML version deleted]]
______________________________________________
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
