[R] lapply to multivariate function?

Rui Barradas ruipbarradas at sapo.pt
Sun Sep 1 23:45:46 CEST 2013


Hello,

I have no experience with packages foreach and doMC.
But I believe that paralel computing only pays if the datasets are 
really large, due to the setup time. Maybe "thousands of observations" 
is not that large.

Rui Barradas

Em 01-09-2013 22:21, Ignacio Martinez escreveu:
> Thanks a lot Rui. Loops make sense to me. I made one modification to your
> code. I have thousands of observation, so I would like to run it in
> parallel. This is my reproducible example:
>
> # Make Data Frame for video actions between given times for user X
> DataVideoActionT <- function (userX, Time1, Time2, Time3){
>    #Get data for user X
>    videoActionsX<-subset(videoLectureActions, username==userX)
>    #Time1 = before first attempt
>    videoActionsX_T1<-subset(videoActionsX, eventTimestamp<Time1)
>    #Time2 = before best attemp
>    videoActionsX_T2<-subset(videoActionsX, eventTimestamp<Time2 &
> eventTimestamp>Time1)
>    #Time3= before last attemp
>    videoActionsX_T3<-subset(videoActionsX, eventTimestamp<Time3 &
> eventTimestamp>Time1)
>
>    error1 = sum(videoActionsX_T1$type==" error ")
>    pause1 = sum(videoActionsX_T1$type==" pause ")
>    play1 = sum(videoActionsX_T1$type==" play ")
>    ratechange1 = sum(videoActionsX_T1$type==" ratechange ")
>    seeked1 = sum(videoActionsX_T1$type==" seeked ")
>    stalled1 = sum(videoActionsX_T1$type==" stalled ")
>
>    error2 = sum(videoActionsX_T2$type==" error ")
>    pause2 = sum(videoActionsX_T2$type==" pause ")
>    play2 = sum(videoActionsX_T2$type==" play ")
>    ratechange2 = sum(videoActionsX_T2$type==" ratechange ")
>    seeked2 = sum(videoActionsX_T2$type==" seeked ")
>    stalled2 = sum(videoActionsX_T2$type==" stalled ")
>
>    error3 = sum(videoActionsX_T3$type==" error ")
>    pause3 = sum(videoActionsX_T3$type==" pause ")
>    play3 = sum(videoActionsX_T3$type==" play ")
>    ratechange3 = sum(videoActionsX_T3$type==" ratechange ")
>    seeked3 = sum(videoActionsX_T3$type==" seeked ")
>    stalled3 = sum(videoActionsX_T3$type==" stalled ")
>
>    data<-data.frame(anon_ID=userX,
>                     error1 = error1,
>                     pause1 = pause1,
>                     play1 = play1,
>                     ratechange1 = ratechange1,
>                     seeked1=seeked1,
>                     stalled1=stalled1,
>                     error2 = error2,
>                     pause2 = pause2,
>                     play2 = play2,
>                     ratechange2 = ratechange2,
>                     seeked2 =seeked2,
>                     stalled2 = stalled2,
>                     error3 = error3,
>                     pause3 = pause3,
>                     play3 = play3,
>                     ratechange3 = ratechange3,
>                     seeked3 = seeked3,
>                     stalled3 = stalled3)
>    return(data)
> }
>
> videoLectureActions<-structure(list(username = c("exampleID1",
> "exampleID1", "exampleID1",
>                                                   "exampleID2",
> "exampleID2", "exampleID2", "exampleID3", "exampleID3",
>                                                   "exampleID3",
> "exampleID3"), currentTime = c("103.701247", "103.701247",
>
>                    "107.543877", "107.543877", "116.456507", "116.456507",
> "119.987188",
>
>                    "177.816693", "183.417124", "183.417124"), playbackRate =
> c("null",
>
>
>    "null", "null", "null", "null", "null", "null", "null", "null",
>
>
>    "null"), pause = c("true", "false", "true", "false", "true",
>
>
>                       "false", "true", "false", "true", "false"), error =
> c("null",
>
>
>
>   "null", "null", "null", "null", "null", "null", "null", "null",
>
>
>
>   "null"), networkState = c("1", "1", "1", "1", "1", "1", "1",
>
>
>
>                           "1", "1", "1"), readyState = c("4", "4", "4", "4",
> "4", "4",
>
>
>
>                                                          "4", "4", "4",
> "4"), lectureID = c("exampleLectureID1", "exampleLectureID1",
>
>
>
>
>                 "exampleLectureID1", "exampleLectureID1",
> "exampleLectureID1",
>
>
>
>
>                 "exampleLectureID1", "exampleLectureID1",
> "exampleLectureID1",
>
>
>
>
>                 "exampleLectureID1", "exampleLectureID1"), eventTimestamp =
> c("2013-03-04 18:51:49",
>
>
>
>
>
>   "2013-03-04 18:51:50", "2013-03-04 18:51:54", "2013-03-04 18:51:56",
>
>
>
>
>
>   "2013-03-04 18:52:05", "2013-03-04 18:52:07", "2013-03-04 18:52:11",
>
>
>
>
>
>   "2013-03-04 18:59:17", "2013-03-04 18:59:23", "2013-03-04 18:59:31"
>
>
>
>
>                 ), initTimestamp = c("2013-03-04 18:44:15", "2013-03-04
> 18:44:15",
>
>
>
>
>                                      "2013-03-04 18:44:15", "2013-03-04
> 18:44:15", "2013-03-04 18:44:15",
>
>
>
>
>                                      "2013-03-04 18:44:15", "2013-03-04
> 18:44:15", "2013-03-04 18:44:15",
>
>
>
>
>                                      "2013-03-04 18:44:15", "2013-03-04
> 18:44:15"), type = c(" pause ",
>
>
>
>
>
>                  " play ", " pause ", " play ", " pause ", " play ", " pause
> ",
>
>
>
>
>
>                  " play ", " pause ", " play "), prevTime = c("103.701247 ",
> "103.701247 ",
>
>
>
>
>
>                                                               "107.543877 ",
> "107.543877 ", "116.456507 ", "116.456507 ", "119.987188 ",
>
>
>
>
>
>                                                               "177.816693 ",
> "183.417124 ", "183.417124 ")), .Names = c("username",
>
>
>
>
>
>
>                                             "currentTime", "playbackRate",
> "pause", "error", "networkState",
>
>
>
>
>
>
>                                             "readyState", "lectureID",
> "eventTimestamp", "initTimestamp",
>
>
>
>
>
>
>                                             "type", "prevTime"), row.names =
> c(1L, 2L, 5L, 6L, 17L, 21L,
>
>
>
>
>
>
>
>    28L, 936L, 957L, 988L), class = "data.frame")
> data<-structure(list(anon_ID = c("exampleID1", "exampleID2", "exampleID3"
> ), maxGrade = c(10, 5, 10), firstGrade = c(10, 5, 8), lastGrade = c(10, 5,
> 10), total_submissions = c(1L, 1L, 3L), Time1 = structure(c(1361993741,
> 1362356090, 1362357401), class = c("POSIXct", "POSIXt"), tzone = ""), TimeM
> = structure(c(1361993741, 1362356090, 1362492744), class = c("POSIXct",
> "POSIXt"), tzone = ""), TimeL = structure(c(1361993741, 1362356090,
> 1362492744), class = c("POSIXct", "POSIXt"), tzone = "")), .Names =
> c("anon_ID", "maxGrade", "firstGrade", "lastGrade", "total_submissions",
> "Time1", "TimeM", "TimeL"), row.names = c(NA, 3L), class = "data.frame")
>
> library(foreach)
> library(doMC)
> registerDoMC(2)  #change the 2 to your number of CPU cores
>
> n <- nrow(data)
> res <- list("vector", n)
> foreach(i=1:n, .verbose=FALSE, .combine=rbind) %do% {
>    res[[i]] <- with(data, DataVideoActionT(anon_ID[i], Time1[i], TimeM[i],
> TimeL[i]))
> }
> test<-do.call(rbind, res)
>
> I have 2 questions.
>
> 1. How can I make foreach not print to the console?
>
> 2. I want to run this in parallel, I i change the %do% for %dopar% the code
> stop working. Instead of getting test with 3 observations and 19 variables
> I get a 2x1 character matrix
>
>
> Thanks!
>
>
>
> On Sun, Sep 1, 2013 at 3:00 PM, Rui Barradas <ruipbarradas at sapo.pt> wrote:
>
>> Hello,
>>
>> Your example doesn't really run, but for what I've seen, if your second
>> data frame is named dat2, something along the lines of
>>
>> n <- nrow(dat2)
>> res <- list("vector", n)
>> for(i in 1:n){
>>          res[[i]] <- with(dat2, DataVideoActionT(anon_ID[i], Time1[i],
>> TimeM[i], TimeL[i]))
>> }
>>
>> do.call(rbind, res)
>>
>>
>> Rui Barradas
>>
>> Em 01-09-2013 17:40, Ignacio Martinez escreveu:
>>
>>> I hope this reproduceble example helps understand what I'm trying to do.
>>>
>>> This is the function:
>>>
>>> # Make Data Frame for video actions between given times for user X
>>> DataVideoActionT <- function (userX, Time1, Time2, Time3){
>>>     #Get data for user X
>>>     videoActionsX<-subset(**videoLectureActions, username==userX)
>>>     #Time1 = before first attempt
>>>     videoActionsX_T1<-subset(**videoActionsX, eventTimestamp<Time1)
>>>     #Time2 = before best attemp
>>>     videoActionsX_T2<-subset(**videoActionsX, eventTimestamp<Time2 &
>>> eventTimestamp>Time1)
>>>     #Time3= before last attemp
>>>     videoActionsX_T3<-subset(**videoActionsX, eventTimestamp<Time3 &
>>> eventTimestamp>Time1)
>>>
>>>     error1 = sum(videoActionsX_T1$type==" error ")
>>>     pause1 = sum(videoActionsX_T1$type==" pause ")
>>>     play1 = sum(videoActionsX_T1$type==" play ")
>>>     ratechange1 = sum(videoActionsX_T1$type==" ratechange ")
>>>     seeked1 = sum(videoActionsX_T1$type==" seeked ")
>>>     stalled1 = sum(videoActionsX_T1$type==" stalled ")
>>>
>>>     error2 = sum(videoActionsX_T2$type==" error ")
>>>     pause2 = sum(videoActionsX_T2$type==" pause ")
>>>     play2 = sum(videoActionsX_T2$type==" play ")
>>>     ratechange2 = sum(videoActionsX_T2$type==" ratechange ")
>>>     seeked2 = sum(videoActionsX_T2$type==" seeked ")
>>>     stalled2 = sum(videoActionsX_T2$type==" stalled ")
>>>
>>>     error3 = sum(videoActionsX_T3$type==" error ")
>>>     pause3 = sum(videoActionsX_T3$type==" pause ")
>>>     play3 = sum(videoActionsX_T3$type==" play ")
>>>     ratechange3 = sum(videoActionsX_T3$type==" ratechange ")
>>>     seeked3 = sum(videoActionsX_T3$type==" seeked ")
>>>     stalled3 = sum(videoActionsX_T3$type==" stalled ")
>>>
>>>     data<-data.frame(anon_ID=**userX,
>>>                      error1 = error1,
>>>                      pause1 = pause1,
>>>                      play1 = play1,
>>>                      ratechange1 = ratechange1,
>>>                      seeked1=seeked1,
>>>                      stalled1=stalled1,
>>>                      error2 = error2,
>>>                      pause2 = pause2,
>>>                      play2 = play2,
>>>                      ratechange2 = ratechange2,
>>>                      seeked2 =seeked2,
>>>                      stalled2 = stalled2,
>>>                      error3 = error3,
>>>                      pause3 = pause3,
>>>                      play3 = play3,
>>>                      ratechange3 = ratechange3,
>>>                      seeked3 = seeked3,
>>>                      stalled3 = stalled3)
>>>     return(data)
>>> }
>>>
>>> This is the videoActionsX  dataframe:
>>>
>>> structure(list(username = c("exampleID1", "exampleID1", "exampleID1",
>>>                               "exampleID2", "exampleID2", "exampleID2",
>>> "exampleID3", "exampleID3",
>>>                               "exampleID3", "exampleID3"), currentTime =
>>> c("103.701247", "103.701247",
>>>
>>>    "107.543877", "107.543877", "116.456507", "116.456507", "119.987188",
>>>
>>>    "177.816693", "183.417124", "183.417124"), playbackRate = c("null",
>>>
>>>                                                            "null", "null",
>>> "null", "null", "null", "null", "null", "null",
>>>
>>>                                                            "null"), pause =
>>> c("true", "false", "true", "false", "true",
>>>
>>>
>>> "false", "true", "false", "true", "false"), error = c("null",
>>>
>>>
>>>                                                         "null", "null",
>>> "null", "null", "null", "null", "null", "null",
>>>
>>>
>>>                                                         "null"),
>>> networkState
>>> = c("1", "1", "1", "1", "1", "1", "1",
>>>
>>>
>>>
>>>       "1", "1", "1"), readyState = c("4", "4", "4", "4", "4", "4",
>>>
>>>
>>>
>>>                                      "4", "4", "4", "4"), lectureID =
>>> c("exampleLectureID1", "exampleLectureID1",
>>>
>>>
>>>
>>>
>>> "exampleLectureID1", "exampleLectureID1", "exampleLectureID1",
>>>
>>>
>>>
>>>
>>> "exampleLectureID1", "exampleLectureID1", "exampleLectureID1",
>>>
>>>
>>>
>>>
>>> "exampleLectureID1", "exampleLectureID1"), eventTimestamp = c("2013-03-04
>>> 18:51:49",
>>>
>>>
>>>
>>>
>>>                                                           "2013-03-04
>>> 18:51:50", "2013-03-04 18:51:54", "2013-03-04 18:51:56",
>>>
>>>
>>>
>>>
>>>                                                           "2013-03-04
>>> 18:52:05", "2013-03-04 18:52:07", "2013-03-04 18:52:11",
>>>
>>>
>>>
>>>
>>>                                                           "2013-03-04
>>> 18:59:17", "2013-03-04 18:59:23", "2013-03-04 18:59:31"
>>>
>>>
>>>
>>>                                                                         ),
>>> initTimestamp = c("2013-03-04 18:44:15", "2013-03-04 18:44:15",
>>>
>>>
>>>
>>>
>>>                  "2013-03-04 18:44:15", "2013-03-04 18:44:15", "2013-03-04
>>> 18:44:15",
>>>
>>>
>>>
>>>
>>>                  "2013-03-04 18:44:15", "2013-03-04 18:44:15", "2013-03-04
>>> 18:44:15",
>>>
>>>
>>>
>>>
>>>                  "2013-03-04 18:44:15", "2013-03-04 18:44:15"), type = c("
>>> pause ",
>>>
>>>
>>>
>>>
>>>                                                                          "
>>> play ", " pause ", " play ", " pause ", " play ", " pause ",
>>>
>>>
>>>
>>>
>>>                                                                          "
>>> play ", " pause ", " play "), prevTime = c("103.701247 ", "103.701247 ",
>>>
>>>
>>>
>>>
>>>
>>>                                           "107.543877 ", "107.543877 ",
>>> "116.456507 ", "116.456507 ", "119.987188 ",
>>>
>>>
>>>
>>>
>>>
>>>                                           "177.816693 ", "183.417124 ",
>>> "183.417124 ")), .Names = c("username",
>>>
>>>
>>>
>>>
>>>
>>>
>>>                         "currentTime", "playbackRate", "pause", "error",
>>> "networkState",
>>>
>>>
>>>
>>>
>>>
>>>
>>>                         "readyState", "lectureID", "eventTimestamp",
>>> "initTimestamp",
>>>
>>>
>>>
>>>
>>>
>>>
>>>                         "type", "prevTime"), row.names = c(1L, 2L, 5L, 6L,
>>> 17L, 21L,
>>>
>>>
>>>
>>>
>>>
>>>
>>>                                                            28L, 936L, 957L,
>>> 988L), class = "data.frame")
>>>
>>>
>>>
>>> But with over 2000 observation.
>>>
>>> And this is the other data frame
>>>
>>> structure(list(anon_ID = c("exampleID1", "exampleID2", "exampleID3" ),
>>> maxGrade = c(10, 5, 10), firstGrade = c(10, 5, 8), lastGrade = c(10,
>>> 5, 10), total_submissions = c(1L, 1L, 3L), Time1 =
>>> structure(c(1361993741, 1362356090, 1362357401), class = c("POSIXct",
>>> "POSIXt"), tzone = ""), TimeM = structure(c(1361993741, 1362356090,
>>> 1362492744), class = c("POSIXct", "POSIXt"), tzone = ""), TimeL =
>>> structure(c(1361993741, 1362356090, 1362492744), class = c("POSIXct",
>>> "POSIXt"), tzone = "")), .Names = c("anon_ID", "maxGrade",
>>> "firstGrade", "lastGrade", "total_submissions", "Time1", "TimeM",
>>> "TimeL"), row.names = c(NA, 3L), class = "data.frame")
>>>
>>>
>>> But with a lot more observations.
>>>
>>>
>>> What I want to do is to call  function (userX, Time1, Time2, Time3)
>>> for all the user in the second data frame where Time1=Time1,
>>> Time2=TimeM, Time3=TimeL
>>>
>>>
>>> I hope that is more clear.
>>>
>>>
>>> Thanks a lot for all the help!
>>>
>>>
>>>
>>> On Sun, Sep 1, 2013 at 11:33 AM, Bert Gunter <gunter.berton at gene.com>
>>> wrote:
>>>
>>>   Oh, another possibility is ?mapply, which I should have pointed out in my
>>>> previous reply. Sorry.
>>>>
>>>> -- Bert
>>>>
>>>>
>>>> On Sun, Sep 1, 2013 at 8:30 AM, Bert Gunter <bgunter at gene.com> wrote:
>>>>
>>>>   Rui et.al.:
>>>>>
>>>>> But apply will not work if the data frame has columns of different
>>>>> classes/types, as appears to be the case here. Viz, from ?apply:
>>>>>
>>>>> "If X is not an array but an object of a class with a non-null dim<
>>>>> http://127.0.0.1:12824/**help/library/base/help/dim<http://127.0.0.1:12824/help/library/base/help/dim>
>>>>>>
>>>>>
>>>>>    value (such as a data frame),apply attempts to coerce it to an array
>>>>> via
>>>>>    as.matrix if it is two-dimensional (e.g., a data frame) or via
>>>>> as.array.
>>>>> "
>>>>>
>>>>> Simply looping by rows (via for() ) appears to be the simplest and
>>>>> probably fastest solution. There are other ways via tapply() and
>>>>> friends,
>>>>> but these are also essentially loops and are likely to incur some
>>>>> additional overhead.
>>>>>
>>>>> All assuming I understand what the OP has requested, of course.
>>>>>
>>>>> Cheers,
>>>>>
>>>>> Bert
>>>>>
>>>>>
>>>>> On Sun, Sep 1, 2013 at 7:31 AM, Rui Barradas <ruipbarradas at sapo.pt
>>>>>> wrote:
>>>>>
>>>>>   Hello,
>>>>>>
>>>>>> Maybe you need apply, not lapply. It seems you want to apply() a
>>>>>> function to the first dimension of your data.frame, something like
>>>>>>
>>>>>> apply(dat, 1, fun)  #apply by rows
>>>>>>
>>>>>>
>>>>>> Hope this helps,
>>>>>>
>>>>>> Rui Barradas
>>>>>>
>>>>>> Em 01-09-2013 15:00, Ignacio Martinez escreveu:
>>>>>>
>>>>>>   I have a Data Frame that contains, between other things, the following
>>>>>>> fields: userX, Time1, Time2, Time3. The number of observations is
>>>>>>> 2000.
>>>>>>>
>>>>>>> I have a function that has as inputs userX, Time1, Time2, Time3 and
>>>>>>> return
>>>>>>> a data frame with 1 observation and 19 variables.
>>>>>>>
>>>>>>> I want to apply that function to all the observations of the first
>>>>>>> data
>>>>>>> frame to make a new data frame with 2000 observations and 19
>>>>>>> variables.
>>>>>>>
>>>>>>> I thought about using lapply, but if I understand correctly, it only
>>>>>>> takes
>>>>>>> one variable.
>>>>>>>
>>>>>>> Can somebody point me in the right direction?
>>>>>>>
>>>>>>> Thanks!
>>>>>>>
>>>>>>>           [[alternative HTML version deleted]]
>>>>>>>
>>>>>>> ______________________________****________________
>>>>>>> R-help at r-project.org mailing list
>>>>>>> https://stat.ethz.ch/mailman/****listinfo/r-help<https://stat.ethz.ch/mailman/**listinfo/r-help>
>>>>>>> <https://stat.**ethz.ch/mailman/listinfo/r-**help<https://stat.ethz.ch/mailman/listinfo/r-help>
>>>>>>>>
>>>>>>> PLEASE do read the posting guide http://www.R-project.org/**
>>>>>>> posting-guide.html <http://www.R-project.org/**posting-guide.html<http://www.R-project.org/posting-guide.html>
>>>>>>>>
>>>>>>>
>>>>>>> and provide commented, minimal, self-contained, reproducible code.
>>>>>>>
>>>>>>>
>>>>>>>   ______________________________****________________
>>>>>> R-help at r-project.org mailing list
>>>>>> https://stat.ethz.ch/mailman/****listinfo/r-help<https://stat.ethz.ch/mailman/**listinfo/r-help>
>>>>>> <https://stat.**ethz.ch/mailman/listinfo/r-**help<https://stat.ethz.ch/mailman/listinfo/r-help>
>>>>>>>
>>>>>> PLEASE do read the posting guide http://www.R-project.org/**
>>>>>> posting-guide.html <http://www.R-project.org/**posting-guide.html<http://www.R-project.org/posting-guide.html>
>>>>>>>
>>>>>>
>>>>>> and provide commented, minimal, self-contained, reproducible code.
>>>>>>
>>>>>>
>>>>>
>>>>>
>>>>> --
>>>>>
>>>>> Bert Gunter
>>>>> Genentech Nonclinical Biostatistics
>>>>>
>>>>> Internal Contact Info:
>>>>> Phone: 467-7374
>>>>> Website:
>>>>>
>>>>> http://pharmadevelopment.**roche.com/index/pdb/pdb-**
>>>>> functional-groups/pdb-**biostatistics/pdb-ncb-home.htm<http://pharmadevelopment.roche.com/index/pdb/pdb-functional-groups/pdb-biostatistics/pdb-ncb-home.htm>
>>>>>
>>>>>
>>>>>
>>>>
>>>>
>>>> --
>>>>
>>>> Bert Gunter
>>>> Genentech Nonclinical Biostatistics
>>>>
>>>> Internal Contact Info:
>>>> Phone: 467-7374
>>>> Website:
>>>>
>>>> http://pharmadevelopment.**roche.com/index/pdb/pdb-**
>>>> functional-groups/pdb-**biostatistics/pdb-ncb-home.htm<http://pharmadevelopment.roche.com/index/pdb/pdb-functional-groups/pdb-biostatistics/pdb-ncb-home.htm>
>>>>
>>>>
>>>>
>>>
>



More information about the R-help mailing list