Fix package, ok for R CMD check - ongoing debug for main function
authorBenjamin Auder <benjamin.auder@somewhere>
Tue, 7 Mar 2017 01:54:03 +0000 (02:54 +0100)
committerBenjamin Auder <benjamin.auder@somewhere>
Tue, 7 Mar 2017 01:58:21 +0000 (02:58 +0100)
epclust/DESCRIPTION
epclust/R/a_NAMESPACE.R [new file with mode: 0644]
epclust/R/clustering.R
epclust/R/de_serialize.R
epclust/R/main.R
epclust/inst/CITATION
epclust/tests/testthat.R
epclust/tests/testthat/test.clustering.R
epclust/tests/testthat/test.de_serialize.R

index 34db956..66d1712 100644 (file)
@@ -12,19 +12,24 @@ Author: Benjamin Auder <Benjamin.Auder@math.u-psud.fr> [aut,cre],
     Jean-Michel Poggi <Jean-Michel.Poggi@math.u-psud.fr> [ctb]
 Maintainer: Benjamin Auder <Benjamin.Auder@math.u-psud.fr>
 Depends:
-    R (>= 3.0.0),
+    R (>= 3.0.0)
+Imports:
+    methods,
     parallel,
     cluster,
-    wavelets
+    wavelets,
+               Rwave
 Suggests:
+    devtools,
     testthat,
                MASS,
                clue,
     wmtsa,
-    RSQLite
+    DBI
 License: MIT + file LICENSE
-RoxygenNote: 5.0.1
+RoxygenNote: 6.0.1
 Collate:
+    'main.R'
     'clustering.R'
     'de_serialize.R'
-    'main.R'
+    'a_NAMESPACE.R'
diff --git a/epclust/R/a_NAMESPACE.R b/epclust/R/a_NAMESPACE.R
new file mode 100644 (file)
index 0000000..3453108
--- /dev/null
@@ -0,0 +1,13 @@
+#' @include de_serialize.R
+#' @include clustering.R
+#' @include main.R
+#'
+#' @importFrom Rwave cwt
+#' @importFrom cluster pam
+#' @importFrom parallel makeCluster clusterExport parLapply stopCluster
+#' @importFrom wavelets dwt wt.filter
+#' @importFrom stats filter spline
+#' @importFrom utils tail
+#' @importFrom methods is
+NULL
+
index 493f90f..6408370 100644 (file)
@@ -1,10 +1,44 @@
-# Cluster one full task (nb_curves / ntasks series); only step 1
-clusteringTask = function(indices, getCoefs, K1, nb_series_per_chunk, ncores)
+#' @name clustering
+#' @rdname clustering
+#' @aliases clusteringTask computeClusters1 computeClusters2
+#'
+#' @title Two-stages clustering, withing one task (see \code{claws()})
+#'
+#' @description \code{clusteringTask()} runs one full task, which consists in iterated stage 1
+#'   clustering (on nb_curves / ntasks energy contributions, computed through discrete
+#'   wavelets coefficients). \code{computeClusters1()} and \code{computeClusters2()}
+#'   correspond to the atomic clustering procedures respectively for stage 1 and 2.
+#'   The former applies the clustering algorithm (PAM) on a contributions matrix, while
+#'   the latter clusters a chunk of series inside one task (~max nb_series_per_chunk)
+#'
+#' @param indices Range of series indices to cluster in parallel (initial data)
+#' @param getContribs Function to retrieve contributions from initial series indices:
+#'   \code{getContribs(indices)} outpus a contributions matrix
+#' @param contribs matrix of contributions (e.g. output of \code{curvesToContribs()})
+#' @inheritParams computeSynchrones
+#' @inheritParams claws
+#'
+#' @return For \code{clusteringTask()} and \code{computeClusters1()}, the indices of the
+#'   computed (K1) medoids. Indices are irrelevant for stage 2 clustering, thus
+#'   \code{computeClusters2()} outputs a matrix of medoids
+#'   (of size limited by nb_series_per_chunk)
+NULL
+
+#' @rdname clustering
+#' @export
+clusteringTask = function(indices, getContribs, K1, nb_series_per_chunk, ncores_clust)
 {
-       cl = parallel::makeCluster(ncores)
-       parallel::clusterExport(cl, varlist=c("getCoefs","K1"), envir=environment())
+
+#NOTE: comment out parallel sections for debugging
+#propagate verbose arg ?!
+
+#      cl = parallel::makeCluster(ncores_clust)
+#      parallel::clusterExport(cl, varlist=c("getContribs","K1"), envir=environment())
        repeat
        {
+
+print(length(indices))
+
                nb_workers = max( 1, floor( length(indices) / nb_series_per_chunk ) )
                indices_workers = lapply( seq_len(nb_workers), function(i)
                        indices[(nb_series_per_chunk*(i-1)+1):(nb_series_per_chunk*i)] )
@@ -16,29 +50,45 @@ clusteringTask = function(indices, getCoefs, K1, nb_series_per_chunk, ncores)
                        indices_workers[[index]] = c(indices_workers[[index]], tail(indices,rem))
                        rem = rem - 1
                }
-               indices = unlist( parallel::parLapply( cl, indices_workers, function(inds) {
-                       require("epclust", quietly=TRUE)
-                       inds[ computeClusters1(getCoefs(inds), K1) ]
+#              indices = unlist( parallel::parLapply( cl, indices_workers, function(inds) {
+               indices = unlist( lapply( indices_workers, function(inds) {
+#                      require("epclust", quietly=TRUE)
+
+print(paste("   ",length(inds))) ## PROBLEME ICI : 21104 ??!
+
+                       inds[ computeClusters1(getContribs(inds), K1) ]
                } ) )
                if (length(indices) == K1)
                        break
        }
-       parallel::stopCluster(cl)
+#      parallel::stopCluster(cl)
        indices #medoids
 }
 
-# Apply the clustering algorithm (PAM) on a coeffs or distances matrix
-computeClusters1 = function(coefs, K1)
-       cluster::pam(coefs, K1, diss=FALSE)$id.med
+#' @rdname clustering
+#' @export
+computeClusters1 = function(contribs, K1)
+       cluster::pam(contribs, K1, diss=FALSE)$id.med
 
-# Cluster a chunk of series inside one task (~max nb_series_per_chunk)
+#' @rdname clustering
+#' @export
 computeClusters2 = function(medoids, K2, getRefSeries, nb_series_per_chunk)
 {
        synchrones = computeSynchrones(medoids, getRefSeries, nb_series_per_chunk)
        medoids[ cluster::pam(computeWerDists(synchrones), K2, diss=TRUE)$medoids , ]
 }
 
-# Compute the synchrones curves (sum of clusters elements) from a clustering result
+#' computeSynchrones
+#'
+#' Compute the synchrones curves (sum of clusters elements) from a matrix of medoids,
+#' using L2 distances.
+#'
+#' @param medoids Matrix of medoids (curves of same legnth as initial series)
+#' @param getRefSeries Function to retrieve initial series (e.g. in stage 2 after series
+#'   have been replaced by stage-1 medoids)
+#' @inheritParams claws
+#'
+#' @export
 computeSynchrones = function(medoids, getRefSeries, nb_series_per_chunk)
 {
        K = nrow(medoids)
@@ -66,16 +116,22 @@ computeSynchrones = function(medoids, getRefSeries, nb_series_per_chunk)
        synchrones[ sapply(seq_len(K), function(i) all(!is.nan(synchrones[i,]))) , ]
 }
 
-# Compute the WER distance between the synchrones curves (in rows)
-computeWerDists = function(curves)
+#' computeWerDists
+#'
+#' Compute the WER distances between the synchrones curves (in rows), which are
+#' returned (e.g.) by \code{computeSynchrones()}
+#'
+#' @param synchrones A matrix of synchrones, in rows. The series have same length as the
+#' series in the initial dataset
+#'
+#' @export
+computeWerDists = function(synchrones)
 {
-       if (!require("Rwave", quietly=TRUE))
-               stop("Unable to load Rwave library")
-       n <- nrow(curves)
-       delta <- ncol(curves)
+       n <- nrow(synchrones)
+       delta <- ncol(synchrones)
        #TODO: automatic tune of all these parameters ? (for other users)
        nvoice   <- 4
-       # noctave = 2^13 = 8192 half hours ~ 180 days ; ~log2(ncol(curves))
+       # noctave = 2^13 = 8192 half hours ~ 180 days ; ~log2(ncol(synchrones))
        noctave = 13
        # 4 here represent 2^5 = 32 half-hours ~ 1 day
        #NOTE: default scalevector == 2^(0:(noctave * nvoice) / nvoice) * s0 (?)
@@ -89,7 +145,7 @@ computeWerDists = function(curves)
 
        # (normalized) observations node with CWT
        Xcwt4 <- lapply(seq_len(n), function(i) {
-               ts <- scale(ts(curves[i,]), center=TRUE, scale=scaled)
+               ts <- scale(ts(synchrones[i,]), center=TRUE, scale=scaled)
                totts.cwt = Rwave::cwt(ts,totnoct,nvoice,w0,plot=0)
                ts.cwt = totts.cwt[,s0log:(s0log+noctave*nvoice)]
                #Normalization
index 242e23a..8dde258 100644 (file)
@@ -1,10 +1,32 @@
-#data: matrix of double or connection
-serialize = function(data_ascii, data_bin_file, nb_per_chunk,
+#' @name de_serialize
+#' @rdname de_serialize
+#' @aliases binarize getDataInFile
+#'
+#' @title (De)Serialization of a matrix
+#'
+#' @description \code{binarize()} serializes a matrix or CSV file with minimal overhead,
+#'   into a binary file. \code{getDataInFile()} achieves the inverse task: she retrieves
+#'   (ASCII) data rows from indices in the binary file
+#'
+#' @param data_ascii Either a matrix or CSV file, with items in rows
+#' @param indices Indices of the lines to retrieve
+#' @param data_bin_file Name of binary file on output (\code{binarize})
+#'   or intput (\code{getDataInFile})
+#' @param nb_per_chunk Number of lines to process in one batch
+#' @inheritParams claws
+#'
+#' @return For \code{getDataInFile()}, the matrix with rows corresponding to the
+#'   requested indices
+NULL
+
+#' @rdname de_serialize
+#' @export
+binarize = function(data_ascii, data_bin_file, nb_per_chunk,
        sep=",", nbytes=4, endian=.Platform$endian)
 {
        if (is.character(data_ascii))
                data_ascii = file(data_ascii, open="r")
-       else if (is(data_ascii,"connection") && !isOpen(data_ascii))
+       else if (methods::is(data_ascii,"connection") && !isOpen(data_ascii))
                open(data_ascii)
 
        first_write = (!file.exists(data_bin_file) || file.info(data_bin_file)$size == 0)
@@ -14,13 +36,13 @@ serialize = function(data_ascii, data_bin_file, nb_per_chunk,
        if (first_write)
        {
                #number of items always on 8 bytes
-               writeBin(0L, data_bin, size=8) #,endian="little")
+               writeBin(0L, data_bin, size=8, endian=endian)
                if (is.matrix(data_ascii))
                        data_length = ncol(data_ascii)
                else #if (is(data, "connection"))
                {
                        data_line = scan(data_ascii, double(), sep=sep, nlines=1, quiet=TRUE)
-                       writeBin(data_line, data_bin, size=nbytes)
+                       writeBin(data_line, data_bin, size=nbytes, endian=endian)
                        data_length = length(data_line)
                }
        }
@@ -43,34 +65,35 @@ serialize = function(data_ascii, data_bin_file, nb_per_chunk,
                        data_chunk = scan(data_ascii, double(), sep=sep, nlines=nb_per_chunk, quiet=TRUE)
                if (length(data_chunk)==0)
                        break
-               writeBin(data_chunk, data_bin, size=nbytes)
+               writeBin(data_chunk, data_bin, size=nbytes, endian=endian)
        }
 
        if (first_write)
        {
                #ecrire file_size-1 / (nbytes*nbWritten) en 0 dans bin_data ! ignored == file_size
                ignored = seek(data_bin, 0)
-               writeBin(data_length, data_bin, size=8)
+               writeBin(data_length, data_bin, size=8, endian=endian)
        }
        close(data_bin)
 
-       if (is(data_ascii,"connection"))
+       if (methods::is(data_ascii,"connection"))
                close(data_ascii)
 }
 
-#read in binary file, always same structure
+#' @rdname de_serialize
+#' @export
 getDataInFile = function(indices, data_bin_file, nbytes=4, endian=.Platform$endian)
 {
        data_bin = file(data_bin_file, "rb")
-       data_size = file.info(data_bin)$size
-       data_length = readBin(data_bin, "integer", 1, 8, endian)
+       data_size = file.info(data_bin_file)$size
+       data_length = readBin(data_bin, "integer", n=1, size=8, endian=endian)
        #Ou t(sapply(...)) (+ rapide ?)
        data_ascii = do.call( rbind, lapply( indices, function(i) {
                offset = 8+(i-1)*data_length*nbytes
                if (offset > data_size)
                        return (vector("double",0))
                ignored = seek(data_bin, offset)
-               readBin(data_bin, "double", n=data_length, size=nbytes)
+               readBin(data_bin, "double", n=data_length, size=nbytes, endian=endian)
        } ) )
        close(data_bin)
        if (ncol(data_ascii)>0) data_ascii else NULL
index 1347fae..b09e934 100644 (file)
@@ -1,7 +1,3 @@
-#' @include de_serialize.R
-#' @include clustering.R
-NULL
-
 #' CLAWS: CLustering with wAvelets and Wer distanceS
 #'
 #' Groups electricity power curves (or any series of similar nature) by applying PAM
@@ -16,21 +12,24 @@ NULL
 #'     \item function: a custom way to retrieve the curves; it has only one argument:
 #'       the indices of the series to be retrieved. See examples
 #'   }
+#' @inheritParams clustering
 #' @param K1 Number of super-consumers to be found after stage 1 (K1 << N)
 #' @param K2 Number of clusters to be found after stage 2 (K2 << K1)
-#' @param random TRUE (default) for random chunks repartition
-#' @param wf Wavelet transform filter; see ?wavelets::wt.filter. Default: haar
+#' @param wf Wavelet transform filter; see ?wavelets::wt.filter
+#' @param ctype Type of contribution: "relative" or "absolute" (or any prefix)
 #' @param WER "end" to apply stage 2 after stage 1 has fully iterated, or "mix" to apply stage 2
 #'   at the end of each task
+#' @param random TRUE (default) for random chunks repartition
 #' @param ntasks Number of tasks (parallel iterations to obtain K1 medoids); default: 1.
 #'   Note: ntasks << N, so that N is "roughly divisible" by N (number of series)
 #' @param ncores_tasks "MPI" number of parallel tasks (1 to disable: sequential tasks)
 #' @param ncores_clust "OpenMP" number of parallel clusterings in one task
 #' @param nb_series_per_chunk (~Maximum) number of series in each group, inside a task
 #' @param min_series_per_chunk Minimum number of series in each group
-#' @param sep Separator in CSV input file (relevant only if getSeries is a file name)
+#' @param sep Separator in CSV input file (if any provided)
 #' @param nbytes Number of bytes to serialize a floating-point number; 4 or 8
 #' @param endian Endianness to use for (de)serialization. Use "little" or "big" for portability
+#' @param verbose Level of verbosity (0/FALSE for nothing or 1/TRUE for all; devel stage)
 #'
 #' @return A matrix of the final medoids curves (K2) in rows
 #'
@@ -42,25 +41,25 @@ NULL
 #' x = seq(0,500,0.05)
 #' L = length(x) #10001
 #' ref_series = matrix( c(cos(x), cos(2*x), cos(3*x), sin(x), sin(2*x), sin(3*x)),
-#'   byrows=TRUE, ncol=L )
+#'   byrow=TRUE, ncol=L )
 #' library(wmtsa)
 #' series = do.call( rbind, lapply( 1:6, function(i)
 #'   do.call(rbind, wmtsa::wavBootstrap(ref_series[i,], n.realization=400)) ) )
 #' #dim(series) #c(2400,10001)
-#' medoids_ascii = claws(series_RData, K1=60, K2=6, wf="d8", nb_series_per_chunk=500)
+#' medoids_ascii = claws(series, K1=60, K2=6, "d8", "rel", nb_series_per_chunk=500)
 #'
 #' # Same example, from CSV file
 #' csv_file = "/tmp/epclust_series.csv"
 #' write.table(series, csv_file, sep=",", row.names=FALSE, col.names=FALSE)
-#' medoids_csv = claws(csv_file, K1=60, K2=6, wf="d8", nb_series_per_chunk=500)
+#' medoids_csv = claws(csv_file, K1=60, K2=6, "d8", "rel", nb_series_per_chunk=500)
 #'
 #' # Same example, from binary file
 #' bin_file = "/tmp/epclust_series.bin"
 #' nbytes = 8
 #' endian = "little"
-#' epclust::serialize(csv_file, bin_file, 500, nbytes, endian)
+#' epclust::binarize(csv_file, bin_file, 500, nbytes, endian)
 #' getSeries = function(indices) getDataInFile(indices, bin_file, nbytes, endian)
-#' medoids_bin = claws(getSeries, K1=60, K2=6, wf="d8", nb_series_per_chunk=500)
+#' medoids_bin = claws(getSeries, K1=60, K2=6, "d8", "rel", nb_series_per_chunk=500)
 #' unlink(csv_file)
 #' unlink(bin_file)
 #'
@@ -69,50 +68,47 @@ NULL
 #' series_db <- dbConnect(RSQLite::SQLite(), "file::memory:")
 #' # Prepare data.frame in DB-format
 #' n = nrow(series)
-#' formatted_series = data.frame(
-#'   ID = rep(1:n,each=L),
-#'   time = as.POSIXct(1800*(0:n),"GMT",origin="2001-01-01"),
-#'   value
-
-
-
-
-#' TODO
-
-
-#' times_values = as.data.frame(series)
+#' time_values = data.frame(
+#'   id = rep(1:n,each=L),
+#'   time = rep( as.POSIXct(1800*(0:n),"GMT",origin="2001-01-01"), L ),
+#'   value = as.double(t(series)) )
 #' dbWriteTable(series_db, "times_values", times_values)
-#' # NOTE: assume that DB internal data is not reorganized when computing coefficients
-#' indexToID_inDB <<- list()
+#' # Fill associative array, map index to identifier
+#' indexToID_inDB <- as.character(
+#'   dbGetQuery(series_db, 'SELECT DISTINCT id FROM time_values')[,"id"] )
 #' getSeries = function(indices) {
-#'   con = dbConnect(drv = RSQLite::SQLite(), dbname = db_file)
-#'   if (indices %in% indexToID_inDB)
-#'   {
-#'     df = dbGetQuery(con, paste(
-#'     "SELECT value FROM times_values GROUP BY id OFFSET ",start,
-#'     "LIMIT ", n, " ORDER BY date", sep=""))
-#'     return (df)
-#'   }
-#'   else
-#'   {
-#'     ...
-#'   }
+#'   request = "SELECT id,value FROM times_values WHERE id in ("
+#'   for (i in indices)
+#'     request = paste(request, i, ",", sep="")
+#'   request = paste(request, ")", sep="")
+#'   df_series = dbGetQuery(series_db, request)
+#'   # Assume that all series share same length at this stage
+#'   ts_length = sum(df_series[,"id"] == df_series[1,"id"])
+#'   t( as.matrix(df_series[,"value"], nrow=ts_length) )
 #' }
-#' dbDisconnect(mydb)
+#' medoids_db = claws(getSeries, K1=60, K2=6, "d8", "rel", nb_series_per_chunk=500)
+#' dbDisconnect(series_db)
+#'
+#' # All computed medoids should be the same:
+#' digest::sha1(medoids_ascii)
+#' digest::sha1(medoids_csv)
+#' digest::sha1(medoids_bin)
+#' digest::sha1(medoids_db)
 #' }
 #' @export
 claws = function(getSeries, K1, K2,
-       random=TRUE, #randomize series order?
-       wf="haar", #stage 1
+       wf,ctype, #stage 1
        WER="end", #stage 2
+       random=TRUE, #randomize series order?
        ntasks=1, ncores_tasks=1, ncores_clust=4, #control parallelism
        nb_series_per_chunk=50*K1, min_series_per_chunk=5*K1, #chunk size
        sep=",", #ASCII input separator
-       nbytes=4, endian=.Platform$endian) #serialization (write,read)
+       nbytes=4, endian=.Platform$endian, #serialization (write,read)
+       verbose=FALSE)
 {
        # Check/transform arguments
        if (!is.matrix(getSeries) && !is.function(getSeries) &&
-               !is(getSeries, "connection" && !is.character(getSeries)))
+               !methods::is(getSeries, "connection" && !is.character(getSeries)))
        {
                stop("'getSeries': matrix, function, file or valid connection (no NA)")
        }
@@ -121,7 +117,7 @@ claws = function(getSeries, K1, K2,
        if (!is.logical(random))
                stop("'random': logical")
        tryCatch(
-               {ignored <- wt.filter(wf)},
+               {ignored <- wavelets::wt.filter(wf)},
                error = function(e) stop("Invalid wavelet filter; see ?wavelets::wt.filter"))
        if (WER!="end" && WER!="mix")
                stop("WER takes values in {'end','mix'}")
@@ -135,30 +131,34 @@ claws = function(getSeries, K1, K2,
        nbytes = .toInteger(nbytes, function(x) x==4 || x==8)
 
        # Serialize series if required, to always use a function
-       bin_dir = "epclust.bin/"
+       bin_dir = ".epclust.bin/"
        dir.create(bin_dir, showWarnings=FALSE, mode="0755")
        if (!is.function(getSeries))
        {
+               if (verbose)
+                       cat("...Serialize time-series\n")
                series_file = paste(bin_dir,"data",sep="") ; unlink(series_file)
-               serialize(getSeries, series_file, nb_series_per_chunk, sep, nbytes, endian)
-               getSeries = function(indices) getDataInFile(indices, series_file, nbytes, endian)
+               binarize(getSeries, series_file, nb_series_per_chunk, sep, nbytes, endian)
+               getSeries = function(inds) getDataInFile(inds, series_file, nbytes, endian)
        }
 
-       # Serialize all wavelets coefficients (+ IDs) onto a file
-       coefs_file = paste(bin_dir,"coefs",sep="") ; unlink(coefs_file)
+       # Serialize all computed wavelets contributions onto a file
+       contribs_file = paste(bin_dir,"contribs",sep="") ; unlink(contribs_file)
        index = 1
        nb_curves = 0
+       if (verbose)
+               cat("...Compute contributions and serialize them\n")
        repeat
        {
                series = getSeries((index-1)+seq_len(nb_series_per_chunk))
                if (is.null(series))
                        break
-               coefs_chunk = curvesToCoefs(series, wf)
-               serialize(coefs_chunk, coefs_file, nb_series_per_chunk, sep, nbytes, endian)
+               contribs_chunk = curvesToContribs(series, wf, ctype)
+               binarize(contribs_chunk, contribs_file, nb_series_per_chunk, sep, nbytes, endian)
                index = index + nb_series_per_chunk
-               nb_curves = nb_curves + nrow(coefs_chunk)
+               nb_curves = nb_curves + nrow(contribs_chunk)
        }
-       getCoefs = function(indices) getDataInFile(indices, coefs_file, nbytes, endian)
+       getContribs = function(indices) getDataInFile(indices, contribs_file, nbytes, endian)
 
        if (nb_curves < min_series_per_chunk)
                stop("Not enough data: less rows than min_series_per_chunk!")
@@ -166,27 +166,36 @@ claws = function(getSeries, K1, K2,
        if (nb_series_per_task < min_series_per_chunk)
                stop("Too many tasks: less series in one task than min_series_per_chunk!")
 
-       # Cluster coefficients in parallel (by nb_series_per_chunk)
+       # Cluster contributions in parallel (by nb_series_per_chunk)
        indices_all = if (random) sample(nb_curves) else seq_len(nb_curves)
        indices_tasks = lapply(seq_len(ntasks), function(i) {
                upper_bound = ifelse( i<ntasks, min(nb_series_per_task*i,nb_curves), nb_curves )
                indices_all[((i-1)*nb_series_per_task+1):upper_bound]
        })
-       cl = parallel::makeCluster(ncores_tasks)
+       if (verbose)
+               cat(paste("...Run ",ntasks," x stage 1 in parallel\n",sep=""))
+#      cl = parallel::makeCluster(ncores_tasks)
+#      parallel::clusterExport(cl, varlist=c("getSeries","getContribs","K1","K2",
+#              "nb_series_per_chunk","ncores_clust","synchrones_file","sep","nbytes","endian"),
+#              envir = environment())
        # 1000*K1 indices [if WER=="end"], or empty vector [if WER=="mix"] --> series on file
-       indices = unlist( parallel::parLapply(cl, indices_tasks, function(inds) {
-               require("epclust", quietly=TRUE)
-               indices_medoids = clusteringTask(inds,getCoefs,K1,nb_series_per_chunk,ncores_clust)
+#      indices = unlist( parallel::parLapply(cl, indices_tasks, function(inds) {
+       indices = unlist( lapply(indices_tasks, function(inds) {
+#              require("epclust", quietly=TRUE)
+
+               browser() #TODO: CONTINUE DEBUG HERE
+
+               indices_medoids = clusteringTask(inds,getContribs,K1,nb_series_per_chunk,ncores_clust)
                if (WER=="mix")
                {
                        medoids2 = computeClusters2(
                                getSeries(indices_medoids), K2, getSeries, nb_series_per_chunk)
-                       serialize(medoids2, synchrones_file, nb_series_per_chunk, sep, nbytes, endian)
+                       binarize(medoids2, synchrones_file, nb_series_per_chunk, sep, nbytes, endian)
                        return (vector("integer",0))
                }
                indices_medoids
        }) )
-       parallel::stopCluster(cl)
+#      parallel::stopCluster(cl)
 
        getRefSeries = getSeries
        synchrones_file = paste(bin_dir,"synchrones",sep="") ; unlink(synchrones_file)
@@ -195,40 +204,62 @@ claws = function(getSeries, K1, K2,
                indices = seq_len(ntasks*K2)
                #Now series must be retrieved from synchrones_file
                getSeries = function(inds) getDataInFile(inds, synchrones_file, nbytes, endian)
-               #Coefs must be re-computed
-               unlink(coefs_file)
+               #Contributions must be re-computed
+               unlink(contribs_file)
                index = 1
+               if (verbose)
+                       cat("...Serialize contributions computed on synchrones\n")
                repeat
                {
                        series = getSeries((index-1)+seq_len(nb_series_per_chunk))
                        if (is.null(series))
                                break
-                       coefs_chunk = curvesToCoefs(series, wf)
-                       serialize(coefs_chunk, coefs_file, nb_series_per_chunk, sep, nbytes, endian)
+                       contribs_chunk = curvesToContribs(series, wf, ctype)
+                       binarize(contribs_chunk, contribs_file, nb_series_per_chunk, sep, nbytes, endian)
                        index = index + nb_series_per_chunk
                }
        }
 
        # Run step2 on resulting indices or series (from file)
+       if (verbose)
+               cat("...Run final // stage 1 + stage 2\n")
        indices_medoids = clusteringTask(
-               indices, getCoefs, K1, nb_series_per_chunk, ncores_tasks*ncores_clust)
-       computeClusters2(getSeries(indices_medoids),K2,getRefSeries,nb_series_per_chunk)
+               indices, getContribs, K1, nb_series_per_chunk, ncores_tasks*ncores_clust)
+       medoids = computeClusters2(getSeries(indices_medoids),K2,getRefSeries,nb_series_per_chunk)
+
+       # Cleanup
+       unlink(bin_dir, recursive=TRUE)
+
+       medoids
 }
 
-# helper
-curvesToCoefs = function(series, wf)
+#' curvesToContribs
+#'
+#' Compute the discrete wavelet coefficients for each series, and aggregate them in
+#' energy contribution across scales as described in https://arxiv.org/abs/1101.4744v2
+#'
+#' @param series Matrix of series (in rows), of size n x L
+#' @inheritParams claws
+#'
+#' @return A matrix of size n x log(L) containing contributions in rows
+#'
+#' @export
+curvesToContribs = function(series, wf, ctype)
 {
        L = length(series[1,])
        D = ceiling( log2(L) )
        nb_sample_points = 2^D
+       cont_types = c("relative","absolute")
+       ctype = cont_types[ pmatch(ctype,cont_types) ]
        t( apply(series, 1, function(x) {
                interpolated_curve = spline(1:L, x, n=nb_sample_points)$y
                W = wavelets::dwt(interpolated_curve, filter=wf, D)@W
-               rev( sapply( W, function(v) ( sqrt( sum(v^2) ) ) ) )
+               nrj = rev( sapply( W, function(v) ( sqrt( sum(v^2) ) ) ) )
+               if (ctype=="relative") nrj / sum(nrj) else nrj
        }) )
 }
 
-# helper
+# Helper for main function: check integer arguments with functiional conditions
 .toInteger <- function(x, condition)
 {
        if (!is.integer(x))
index 23e7819..b6f2f6d 100644 (file)
@@ -4,8 +4,8 @@ citEntry(entry = "Manual",
   title        = ".",
   author       = personList(as.person("Benjamin Auder"),
                    as.person("Jairo Cugliari"),
-                   as.person("Yannig Goude")),
-                   as.person("Jean-Michel Poggi"))
+                   as.person("Yannig Goude"),
+                   as.person("Jean-Michel Poggi")),
   organization = "Paris-Sud, Saclay & Lyon 2",
   address      = "Orsay, Saclay & Lyon, France",
   year         = "2017",
index f58c71c..eb8bc36 100644 (file)
@@ -1,4 +1,4 @@
 library(testthat)
-load_all("..")
+library(epclust)
 
 test_check("epclust")
index b6231e2..9333876 100644 (file)
@@ -7,7 +7,8 @@ I = function(i, base)
 test_that("computeClusters1 behave as expected",
 {
        require("MASS", quietly=TRUE)
-       library("clue", quietly=TRUE)
+       if (!require("clue", quietly=TRUE))
+               skip("'clue' package not available")
 
        # 3 gaussian clusters, 300 items; and then 7 gaussian clusters, 490 items
        n = 300
@@ -126,8 +127,9 @@ test_that("clusteringTask + computeClusters2 behave as expected",
                if (length(indices)>0) series[indices,] else NULL
        }
        wf = "haar"
-       getCoefs = function(indices) curvesToCoefs(series[indices,],wf)
-       medoids_K1 = getSeries( clusteringTask(1:n, getCoefs, K1, 75, 4) )
+       ctype = "absolute"
+       getContribs = function(indices) curvesToContribs(series[indices,],wf,ctype)
+       medoids_K1 = getSeries( clusteringTask(1:n, getContribs, K1, 75, 4) )
        medoids_K2 = computeClusters2(medoids_K1, K2, getSeries, 120)
 
        expect_equal(dim(medoids_K1), c(K1,L))
index 27e6d59..a2fae5e 100644 (file)
@@ -1,17 +1,17 @@
 context("de_serialize")
 
-data_bin_file <<- "/tmp/epclust_test.bin"
-unlink(data_bin_file)
-
 test_that("serialization + getDataInFile retrieve original data / from matrix",
 {
+       data_bin_file = "/tmp/epclust_test_m.bin"
+       unlink(data_bin_file)
+
        #dataset 200 lignes / 30 columns
        data_ascii = matrix(runif(200*30,-10,10),ncol=30)
        nbytes = 4 #lead to a precision of 1e-7 / 1e-8
        endian = "little"
 
        #Simulate serialization in one single call
-       serialize(data_ascii, data_bin_file, 500, ",", nbytes, endian)
+       binarize(data_ascii, data_bin_file, 500, ",", nbytes, endian)
        expect_equal(file.info(data_bin_file)$size, length(data_ascii)*nbytes+8)
        for (indices in list(c(1,3,5), 3:13, c(5,20,50), c(75,130:135), 196:200))
        {
@@ -22,7 +22,7 @@ test_that("serialization + getDataInFile retrieve original data / from matrix",
 
        #...in several calls (last call complete, next call NULL)
        for (i in 1:20)
-               serialize(data_ascii[((i-1)*10+1):(i*10),], data_bin_file, 20, ",", nbytes, endian)
+               binarize(data_ascii[((i-1)*10+1):(i*10),], data_bin_file, 20, ",", nbytes, endian)
        expect_equal(file.info(data_bin_file)$size, length(data_ascii)*nbytes+8)
        for (indices in list(c(1,3,5), 3:13, c(5,20,50), c(75,130:135), 196:200))
        {
@@ -34,14 +34,17 @@ test_that("serialization + getDataInFile retrieve original data / from matrix",
 
 test_that("serialization + getDataInFile retrieve original data / from connection",
 {
+       data_bin_file = "/tmp/epclust_test_c.bin"
+       unlink(data_bin_file)
+
        #dataset 300 lignes / 50 columns
        data_csv = system.file("testdata","de_serialize.csv",package="epclust")
        nbytes = 8
        endian = "big"
-       data_ascii = as.matrix(read.csv(test_series, sep=";", header=FALSE)) #for ref
+       data_ascii = as.matrix(read.csv(data_csv, sep=";", header=FALSE)) #for ref
 
        #Simulate serialization in one single call
-       serialize(data_csv, data_bin_file, 350, ";", nbytes, endian)
+       binarize(data_csv, data_bin_file, 350, ";", nbytes, endian)
        expect_equal(file.info(data_bin_file)$size, 300*50*8+8)
        for (indices in list(c(1,3,5), 3:13, c(5,20,50), c(75,130:135), 196:200))
        {
@@ -53,7 +56,7 @@ test_that("serialization + getDataInFile retrieve original data / from connectio
 
        #...in several calls / chunks of 29 --> 29*10 + 10, incomplete last
        data_con = file(data_csv, "r")
-       serialize(data_con, data_bin_file, 29, ";", nbytes, endian)
+       binarize(data_con, data_bin_file, 29, ";", nbytes, endian)
        expect_equal(file.info(data_bin_file)$size, 300*50*8+8)
        for (indices in list(c(1,3,5), 3:13, c(5,20,50), c(75,130:135), 196:200))
        {
@@ -61,5 +64,5 @@ test_that("serialization + getDataInFile retrieve original data / from connectio
                expect_equal(data_lines, data_ascii[indices,])
        }
        unlink(data_bin_file)
-       #close(data_con) --> done in serialize()
+       #close(data_con) --> done in binarize()
 })