Hi
I slightly modified Jim's code
first part is function to split data frame test according to act, juln and day
and compute repetitions in each chunk.
fff<- function(x) {
fac <- factor((x[, "act"]==0)*1+(x[,"act"] == 200)*2, levels=c(1,0,2))
int<-interaction(x[,"juln"], x[,"day"], fac)
res <- cumsum(c(1, abs(diff(as.numeric(int)))))
res
}
test$fac<-fff(test)
Second part evaluates length of each chunk
test$res <- ave(test$fac, test$fac, FUN=length)
Last part computes max (min, sum) of res in each distinct chunk.
fff2<- function(x) {
fac <- factor((x[, "act"]==0)*1+(x[,"act"] == 200)*2, levels=c(1,0,2),
labels=c("0", "1-199", "200"))
fac
}
aggregate(test$res, list(test$juln, test$day), max)
aggregate(test$res, list(test$juln, test$day, fff2(test)), max)
Is it what you want?
Petr
From: zuzana zajkova [mailto:zuzu...@gmail.com]
Sent: Friday, May 03, 2013 7:10 PM
To: PIKAL Petr; jholt...@gmail.com
Cc: r-help@r-project.org
Subject: Re: [R] Counting number of consecutive occurrences per rows
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<mailto: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-boun...@r-project.org>
> [mailto:r-help-bounces@r-<mailto:r-help-bounces@r->
> project.org<http://project.org>] On Behalf Of zuzana zajkova
> Sent: Monday, April 29, 2013 12:45 PM
> To: r-help@r-project.org<mailto: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<mailto: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.
