Commit | Line | Data |
---|---|---|
8702eb86 | 1 | #' CLAWS: CLustering with wAvelets and Wer distanceS |
7f0781b7 | 2 | #' |
eef6f6c9 BA |
3 | #' Cluster electricity power curves (or any series of similar nature) by applying a |
4 | #' two stage procedure in parallel (see details). | |
5 | #' Input series must be sampled on the same time grid, no missing values. | |
6 | #' | |
d9bb53c5 BA |
7 | #' Summary of the function execution flow: |
8 | #' \enumerate{ | |
9 | #' \item Compute and serialize all contributions, obtained through discrete wavelet | |
10 | #' decomposition (see Antoniadis & al. [2013]) | |
11 | #' \item Divide series into \code{ntasks} groups to process in parallel. In each task: | |
eef6f6c9 | 12 | #' \enumerate{ |
d9bb53c5 | 13 | #' \item iterate the first clustering algorithm on its aggregated outputs, |
3fb6e823 BA |
14 | #' on inputs of size \code{nb_items_clust}\cr |
15 | #' -> K1 medoids indices | |
16 | #' \item optionally, if WER=="mix":\cr | |
17 | #' a. compute WER distances (K1xK1) between medoids\cr | |
18 | #' b. apply the 2nd clustering algorithm\cr | |
19 | #' -> K2 medoids indices | |
eef6f6c9 | 20 | #' } |
d9bb53c5 | 21 | #' \item Launch a final task on the aggregated outputs of all previous tasks: |
40f12a2f BA |
22 | #' ntasks*K1 if WER=="end", ntasks*K2 otherwise |
23 | #' \item Compute synchrones (sum of series within each final group) | |
d9bb53c5 | 24 | #' } |
3fb6e823 | 25 | #' |
40f12a2f BA |
26 | #' The main argument -- \code{series} -- has a quite misleading name, since it can be |
27 | #' either a [big.]matrix, a CSV file, a connection or a user function to retrieve series. | |
3fb6e823 BA |
28 | #' When \code{series} is given as a function it must take a single argument, |
29 | #' 'indices': integer vector equal to the indices of the curves to retrieve; | |
40f12a2f | 30 | #' see SQLite example. |
a52836b2 | 31 | #' WARNING: the return value must be a matrix (in columns), or NULL if no matches. |
3fb6e823 | 32 | #' |
d9bb53c5 | 33 | #' Note: Since we don't make assumptions on initial data, there is a possibility that |
40f12a2f | 34 | #' even when serialized, contributions do not fit in RAM. For example, |
d9bb53c5 BA |
35 | #' 30e6 series of length 100,000 would lead to a +4Go contribution matrix. Therefore, |
36 | #' it's safer to place these in (binary) files; that's what we do. | |
7f0781b7 | 37 | #' |
3fb6e823 | 38 | #' @param series Access to the N (time-)series, which can be of one of the four |
8702eb86 BA |
39 | #' following types: |
40 | #' \itemize{ | |
eef6f6c9 | 41 | #' \item [big.]matrix: each column contains the (time-ordered) values of one time-serie |
bf5c0844 BA |
42 | #' \item connection: any R connection object providing lines as described above |
43 | #' \item character: name of a CSV file containing series in rows (no header) | |
8702eb86 | 44 | #' \item function: a custom way to retrieve the curves; it has only one argument: |
eef6f6c9 | 45 | #' the indices of the series to be retrieved. See SQLite example |
8702eb86 | 46 | #' } |
3fb6e823 | 47 | #' @param K1 Number of clusters to be found after stage 1 (K1 << N) |
1c6f223e | 48 | #' @param K2 Number of clusters to be found after stage 2 (K2 << K1) |
3fb6e823 BA |
49 | #' @param nb_series_per_chunk Number of series to retrieve in one batch |
50 | #' @param nb_items_clust Number of items in 1st clustering algorithm input | |
0486fbad | 51 | #' @param algoClust1 Clustering algorithm for stage 1. A function which takes (data, K) |
2b9f5356 | 52 | #' as argument where data is a matrix in columns and K the desired number of clusters, |
3fb6e823 | 53 | #' and outputs K medoids ranks. Default: PAM. |
0486fbad | 54 | #' @param algoClust2 Clustering algorithm for stage 2. A function which takes (dists, K) |
2b9f5356 | 55 | #' as argument where dists is a matrix of distances and K the desired number of clusters, |
3fb6e823 | 56 | #' and outputs K medoids ranks. Default: PAM. |
eef6f6c9 BA |
57 | #' @param wav_filt Wavelet transform filter; see ?wavelets::wt.filter |
58 | #' @param contrib_type Type of contribution: "relative", "logit" or "absolute" (any prefix) | |
59 | #' @param WER "end" to apply stage 2 after stage 1 has fully iterated, or "mix" to apply | |
60 | #' stage 2 at the end of each task | |
3fb6e823 | 61 | #' @param smooth_lvl Smoothing level: odd integer, 1 == no smoothing. |
a52836b2 | 62 | #' @param nvoice Number of voices within each octave for CWT computations |
4bcfdbee | 63 | #' @param random TRUE (default) for random chunks repartition |
eef6f6c9 BA |
64 | #' @param ntasks Number of tasks (parallel iterations to obtain K1 [if WER=="end"] |
65 | #' or K2 [if WER=="mix"] medoids); default: 1. | |
66 | #' Note: ntasks << N (number of series), so that N is "roughly divisible" by ntasks | |
3fb6e823 BA |
67 | #' @param ncores_tasks Number of parallel tasks ('1' == sequential tasks) |
68 | #' @param ncores_clust Number of parallel clusterings in one task | |
4bcfdbee | 69 | #' @param sep Separator in CSV input file (if any provided) |
e0154a59 | 70 | #' @param nbytes 4 or 8 bytes to (de)serialize a floating-point number |
3fb6e823 BA |
71 | #' @param endian Endianness for (de)serialization: "little" or "big" |
72 | #' @param verbose FALSE: nothing printed; TRUE: some execution traces | |
73 | #' @param parll TRUE: run in parallel. FALSE: run sequentially | |
7f0781b7 | 74 | #' |
3fb6e823 | 75 | #' @return A list: |
40f12a2f | 76 | #' \itemize{ |
3fb6e823 BA |
77 | #' \item medoids: matrix of the final K2 medoids curves |
78 | #' \item ranks: corresponding indices in the dataset | |
79 | #' \item synchrones: sum of series within each final group | |
40f12a2f | 80 | #' } |
eef6f6c9 BA |
81 | #' |
82 | #' @references Clustering functional data using Wavelets [2013]; | |
83 | #' A. Antoniadis, X. Brossat, J. Cugliari & J.-M. Poggi. | |
84 | #' Inter. J. of Wavelets, Multiresolution and Information Procesing, | |
85 | #' vol. 11, No 1, pp.1-30. doi:10.1142/S0219691313500033 | |
1c6f223e BA |
86 | #' |
87 | #' @examples | |
4efef8cc | 88 | #' \dontrun{ |
eef6f6c9 | 89 | #' # WER distances computations are too long for CRAN (for now) |
e0154a59 | 90 | #' parll = FALSE #on this small example, sequential run is faster |
4efef8cc BA |
91 | #' |
92 | #' # Random series around cos(x,2x,3x)/sin(x,2x,3x) | |
3fb6e823 BA |
93 | #' x <- seq(0,50,0.05) |
94 | #' L <- length(x) #1001 | |
282342ba | 95 | #' ref_series <- matrix( c(cos(x),cos(2*x),cos(3*x),sin(x),sin(2*x),sin(3*x)), ncol=6 ) |
4efef8cc | 96 | #' library(wmtsa) |
282342ba | 97 | #' series <- do.call( cbind, lapply( 1:6, function(i) |
3fb6e823 | 98 | #' do.call(cbind, wmtsa::wavBootstrap(ref_series[,i], n.realization=40)) ) ) |
dc86eb0c BA |
99 | #' # Mix series so that all groups are evenly spread |
100 | #' permut <- (0:239)%%6 * 40 + (0:239)%/%6 + 1 | |
101 | #' series = series[,permut] | |
3fb6e823 | 102 | #' #dim(series) #c(240,1001) |
e0154a59 BA |
103 | #' res_ascii <- claws(series, K1=30, K2=6, nb_series_per_chunk=500, |
104 | #' nb_items_clust=100, random=FALSE, verbose=TRUE, parll=parll) | |
4efef8cc BA |
105 | #' |
106 | #' # Same example, from CSV file | |
3fb6e823 BA |
107 | #' csv_file <- tempfile(pattern="epclust_series.csv_") |
108 | #' write.table(t(series), csv_file, sep=",", row.names=FALSE, col.names=FALSE) | |
e0154a59 | 109 | #' res_csv <- claws(csv_file, 30, 6, 500, 100, random=FALSE, parll=parll) |
4efef8cc BA |
110 | #' |
111 | #' # Same example, from binary file | |
3fb6e823 | 112 | #' bin_file <- tempfile(pattern="epclust_series.bin_") |
eef6f6c9 BA |
113 | #' nbytes <- 8 |
114 | #' endian <- "little" | |
dc86eb0c | 115 | #' binarize(csv_file, bin_file, 500, ",", nbytes, endian) |
eef6f6c9 | 116 | #' getSeries <- function(indices) getDataInFile(indices, bin_file, nbytes, endian) |
e0154a59 | 117 | #' res_bin <- claws(getSeries, 30, 6, 500, 100, random=FALSE, parll=parll) |
4efef8cc BA |
118 | #' unlink(csv_file) |
119 | #' unlink(bin_file) | |
120 | #' | |
121 | #' # Same example, from SQLite database | |
122 | #' library(DBI) | |
123 | #' series_db <- dbConnect(RSQLite::SQLite(), "file::memory:") | |
124 | #' # Prepare data.frame in DB-format | |
dc86eb0c BA |
125 | #' n <- ncol(series) |
126 | #' times_values <- data.frame( | |
127 | #' id = rep(1:n,each=L), | |
128 | #' time = rep( as.POSIXct(1800*(1:L),"GMT",origin="2001-01-01"), n ), | |
129 | #' value = as.double(series) ) | |
4efef8cc | 130 | #' dbWriteTable(series_db, "times_values", times_values) |
4bcfdbee BA |
131 | #' # Fill associative array, map index to identifier |
132 | #' indexToID_inDB <- as.character( | |
dc86eb0c | 133 | #' dbGetQuery(series_db, 'SELECT DISTINCT id FROM times_values')[,"id"] ) |
eef6f6c9 | 134 | #' serie_length <- as.integer( dbGetQuery(series_db, |
dc86eb0c | 135 | #' paste("SELECT COUNT(*) FROM times_values WHERE id == ",indexToID_inDB[1],sep="")) ) |
eef6f6c9 | 136 | #' getSeries <- function(indices) { |
e0154a59 BA |
137 | #' indices = indices[ indices <= length(indexToID_inDB) ] |
138 | #' if (length(indices) == 0) | |
139 | #' return (NULL) | |
eef6f6c9 | 140 | #' request <- "SELECT id,value FROM times_values WHERE id in (" |
dc86eb0c BA |
141 | #' for (i in seq_along(indices)) { |
142 | #' request <- paste(request, indexToID_inDB[ indices[i] ], sep="") | |
143 | #' if (i < length(indices)) | |
144 | #' request <- paste(request, ",", sep="") | |
145 | #' } | |
eef6f6c9 BA |
146 | #' request <- paste(request, ")", sep="") |
147 | #' df_series <- dbGetQuery(series_db, request) | |
e0154a59 | 148 | #' matrix(df_series[,"value"], nrow=serie_length) |
4efef8cc | 149 | #' } |
e0154a59 | 150 | #' res_db <- claws(getSeries, 30, 6, 500, 100, random=FALSE, parll=parll) |
4bcfdbee BA |
151 | #' dbDisconnect(series_db) |
152 | #' | |
dc86eb0c BA |
153 | #' # All results should be equal: |
154 | #' all(res_ascii$ranks == res_csv$ranks | |
155 | #' & res_ascii$ranks == res_bin$ranks | |
156 | #' & res_ascii$ranks == res_db$ranks) | |
1c6f223e | 157 | #' } |
1c6f223e | 158 | #' @export |
e0154a59 | 159 | claws <- function(series, K1, K2, nb_series_per_chunk, nb_items_clust=5*K1, |
40f12a2f BA |
160 | algoClust1=function(data,K) cluster::pam(t(data),K,diss=FALSE,pamonce=1)$id.med, |
161 | algoClust2=function(dists,K) cluster::pam(dists,K,diss=TRUE,pamonce=1)$id.med, | |
282342ba BA |
162 | wav_filt="d8", contrib_type="absolute", WER="end", smooth_lvl=3, nvoice=4, |
163 | random=TRUE, ntasks=1, ncores_tasks=1, ncores_clust=3, sep=",", nbytes=4, | |
e0154a59 | 164 | endian=.Platform$endian, verbose=FALSE, parll=TRUE) |
ac1d4231 | 165 | { |
0e2dce80 | 166 | # Check/transform arguments |
40f12a2f BA |
167 | if (!is.matrix(series) && !bigmemory::is.big.matrix(series) |
168 | && !is.function(series) | |
169 | && !methods::is(series,"connection") && !is.character(series)) | |
0e2dce80 | 170 | { |
40f12a2f | 171 | stop("'series': [big]matrix, function, file or valid connection (no NA)") |
5c652979 | 172 | } |
eef6f6c9 BA |
173 | K1 <- .toInteger(K1, function(x) x>=2) |
174 | K2 <- .toInteger(K2, function(x) x>=2) | |
37c82bba | 175 | nb_series_per_chunk <- .toInteger(nb_series_per_chunk, function(x) x>=1) |
3c5a4b08 | 176 | nb_items_clust <- .toInteger(nb_items_clust, function(x) x>K1) |
eef6f6c9 | 177 | random <- .toLogical(random) |
282342ba BA |
178 | tryCatch({ignored <- wavelets::wt.filter(wav_filt)}, |
179 | error=function(e) stop("Invalid wavelet filter; see ?wavelets::wt.filter") ) | |
180 | ctypes <- c("relative","absolute","logit") | |
181 | contrib_type <- ctypes[ pmatch(contrib_type,ctypes) ] | |
eef6f6c9 BA |
182 | if (is.na(contrib_type)) |
183 | stop("'contrib_type' in {'relative','absolute','logit'}") | |
7f0781b7 | 184 | if (WER!="end" && WER!="mix") |
eef6f6c9 BA |
185 | stop("'WER': in {'end','mix'}") |
186 | random <- .toLogical(random) | |
187 | ntasks <- .toInteger(ntasks, function(x) x>=1) | |
188 | ncores_tasks <- .toInteger(ncores_tasks, function(x) x>=1) | |
189 | ncores_clust <- .toInteger(ncores_clust, function(x) x>=1) | |
56857861 BA |
190 | if (!is.character(sep)) |
191 | stop("'sep': character") | |
eef6f6c9 BA |
192 | nbytes <- .toInteger(nbytes, function(x) x==4 || x==8) |
193 | verbose <- .toLogical(verbose) | |
194 | parll <- .toLogical(parll) | |
56857861 | 195 | |
40f12a2f | 196 | # Binarize series if it is not a function; the aim is to always use a function, |
2b9f5356 | 197 | # to uniformize treatments. An equally good alternative would be to use a file-backed |
d9bb53c5 | 198 | # bigmemory::big.matrix, but it would break the "all-is-function" pattern. |
40f12a2f | 199 | if (!is.function(series)) |
56857861 | 200 | { |
4bcfdbee | 201 | if (verbose) |
a52836b2 | 202 | cat("...Serialize time-series (or retrieve past binary file)\n") |
282342ba | 203 | series_file <- ".series.epclust.bin" |
e0154a59 | 204 | if (!file.exists(series_file)) |
40f12a2f | 205 | binarize(series, series_file, nb_series_per_chunk, sep, nbytes, endian) |
282342ba | 206 | getSeries <- function(inds) getDataInFile(inds, series_file, nbytes, endian) |
56857861 | 207 | } |
40f12a2f | 208 | else |
282342ba | 209 | getSeries <- series |
ac1d4231 | 210 | |
95b5c2e6 | 211 | # Serialize all computed wavelets contributions into a file |
282342ba | 212 | contribs_file <- ".contribs.epclust.bin" |
4bcfdbee | 213 | if (verbose) |
a52836b2 | 214 | cat("...Compute contributions and serialize them (or retrieve past binary file)\n") |
e0154a59 | 215 | if (!file.exists(contribs_file)) |
a52836b2 | 216 | { |
282342ba | 217 | nb_curves <- binarizeTransform(getSeries, |
40f12a2f | 218 | function(curves) curvesToContribs(curves, wav_filt, contrib_type), |
a52836b2 BA |
219 | contribs_file, nb_series_per_chunk, nbytes, endian) |
220 | } | |
221 | else | |
222 | { | |
223 | # TODO: duplicate from getDataInFile() in de_serialize.R | |
282342ba BA |
224 | contribs_size <- file.info(contribs_file)$size #number of bytes in the file |
225 | contrib_length <- readBin(contribs_file, "integer", n=1, size=8, endian=endian) | |
226 | nb_curves <- (contribs_size-8) / (nbytes*contrib_length) | |
a52836b2 | 227 | } |
282342ba | 228 | getContribs <- function(indices) getDataInFile(indices, contribs_file, nbytes, endian) |
8e6accca | 229 | |
2b9f5356 | 230 | # A few sanity checks: do not continue if too few data available. |
eef6f6c9 BA |
231 | if (nb_curves < K2) |
232 | stop("Not enough data: less series than final number of clusters") | |
282342ba | 233 | nb_series_per_task <- round(nb_curves / ntasks) |
eef6f6c9 BA |
234 | if (nb_series_per_task < K2) |
235 | stop("Too many tasks: less series in one task than final number of clusters") | |
ac1d4231 | 236 | |
d9bb53c5 BA |
237 | # Generate a random permutation of 1:N (if random==TRUE); |
238 | # otherwise just use arrival (storage) order. | |
282342ba | 239 | indices_all <- if (random) sample(nb_curves) else seq_len(nb_curves) |
2b9f5356 | 240 | # Split (all) indices into ntasks groups of ~same size |
282342ba BA |
241 | indices_tasks <- lapply(seq_len(ntasks), function(i) { |
242 | upper_bound <- ifelse( i<ntasks, min(nb_series_per_task*i,nb_curves), nb_curves ) | |
2b9f5356 BA |
243 | indices_all[((i-1)*nb_series_per_task+1):upper_bound] |
244 | }) | |
245 | ||
246 | if (parll && ntasks>1) | |
247 | { | |
248 | # Initialize parallel runs: outfile="" allow to output verbose traces in the console | |
249 | # under Linux. All necessary variables are passed to the workers. | |
3fb6e823 BA |
250 | cl <- |
251 | if (verbose) | |
252 | parallel::makeCluster(ncores_tasks, outfile="") | |
253 | else | |
254 | parallel::makeCluster(ncores_tasks) | |
282342ba | 255 | varlist <- c("ncores_clust","verbose","parll", #task 1 & 2 |
3c5a4b08 BA |
256 | "K1","getContribs","algoClust1","nb_items_clust") #task 1 |
257 | if (WER=="mix") | |
258 | { | |
259 | # Add variables for task 2 | |
282342ba BA |
260 | varlist <- c(varlist, "K2","getSeries","algoClust2","nb_series_per_chunk", |
261 | "smooth_lvl","nvoice","nbytes","endian") | |
3c5a4b08 | 262 | } |
282342ba | 263 | parallel::clusterExport(cl, varlist, envir <- environment()) |
2b9f5356 BA |
264 | } |
265 | ||
266 | # This function achieves one complete clustering task, divided in stage 1 + stage 2. | |
3c5a4b08 BA |
267 | # stage 1: n indices --> clusteringTask1(...) --> K1 medoids (indices) |
268 | # stage 2: K1 indices --> K1xK1 WER distances --> clusteringTask2(...) --> K2 medoids, | |
282342ba BA |
269 | # where n == N / ntasks, N being the total number of curves. |
270 | runTwoStepClustering <- function(inds) | |
492cd9e7 | 271 | { |
d9bb53c5 | 272 | # When running in parallel, the environment is blank: we need to load the required |
2b9f5356 | 273 | # packages, and pass useful variables. |
bf5c0844 | 274 | if (parll && ntasks>1) |
492cd9e7 | 275 | require("epclust", quietly=TRUE) |
282342ba | 276 | indices_medoids <- clusteringTask1(inds, getContribs, K1, algoClust1, |
3c5a4b08 | 277 | nb_items_clust, ncores_clust, verbose, parll) |
40f12a2f | 278 | if (WER=="mix") |
56857861 | 279 | { |
282342ba BA |
280 | indices_medoids <- clusteringTask2(indices_medoids, getSeries, K2, algoClust2, |
281 | nb_series_per_chunk,smooth_lvl,nvoice,nbytes,endian,ncores_clust,verbose,parll) | |
56857861 BA |
282 | } |
283 | indices_medoids | |
492cd9e7 BA |
284 | } |
285 | ||
c45fd663 | 286 | if (verbose) |
e161499b | 287 | { |
282342ba | 288 | message <- paste("...Run ",ntasks," x stage 1", sep="") |
e161499b | 289 | if (WER=="mix") |
282342ba | 290 | message <- paste(message," + stage 2", sep="") |
e161499b BA |
291 | cat(paste(message,"\n", sep="")) |
292 | } | |
c45fd663 | 293 | |
3c5a4b08 BA |
294 | # As explained above, we obtain after all runs ntasks*[K1 or K2] medoids indices, |
295 | # depending wether WER=="end" or "mix", respectively. | |
40f12a2f | 296 | indices_medoids_all <- |
eef6f6c9 BA |
297 | if (parll && ntasks>1) |
298 | unlist( parallel::parLapply(cl, indices_tasks, runTwoStepClustering) ) | |
299 | else | |
300 | unlist( lapply(indices_tasks, runTwoStepClustering) ) | |
40f12a2f | 301 | |
bf5c0844 | 302 | if (parll && ntasks>1) |
492cd9e7 | 303 | parallel::stopCluster(cl) |
3465b246 | 304 | |
3c5a4b08 BA |
305 | # For the last stage, ncores_tasks*(ncores_clusts+1) cores should be available: |
306 | # - ntasks for level 1 parallelism | |
307 | # - ntasks*ncores_clust for level 2 parallelism, | |
308 | # but since an extension MPI <--> tasks / OpenMP <--> sub-tasks is on the way, | |
309 | # it's better to just re-use ncores_clust | |
310 | ncores_last_stage <- ncores_clust | |
0e2dce80 | 311 | |
3c5a4b08 | 312 | # Run last clustering tasks to obtain only K2 medoids indices |
4bcfdbee BA |
313 | if (verbose) |
314 | cat("...Run final // stage 1 + stage 2\n") | |
282342ba | 315 | indices_medoids <- clusteringTask1(indices_medoids_all, getContribs, K1, algoClust1, |
3c5a4b08 | 316 | nb_items_clust, ncores_tasks*ncores_clust, verbose, parll) |
dc86eb0c BA |
317 | |
318 | indices_medoids <- clusteringTask2(indices_medoids, getSeries, K2, algoClust2, | |
282342ba | 319 | nb_series_per_chunk,smooth_lvl,nvoice,nbytes,endian,ncores_last_stage,verbose,parll) |
4bcfdbee | 320 | |
3c5a4b08 BA |
321 | # Compute synchrones, that is to say the cumulated power consumptions for each of the K2 |
322 | # final groups. | |
282342ba BA |
323 | medoids <- getSeries(indices_medoids) |
324 | synchrones <- computeSynchrones(medoids, getSeries, nb_curves, nb_series_per_chunk, | |
3c5a4b08 | 325 | ncores_last_stage, verbose, parll) |
4bcfdbee | 326 | |
40f12a2f BA |
327 | # NOTE: no need to use big.matrix here, since there are only K2 << K1 << N remaining curves |
328 | list("medoids"=medoids, "ranks"=indices_medoids, "synchrones"=synchrones) | |
cea14f3a | 329 | } |