From: Benjamin Auder Date: Sat, 18 Mar 2017 01:24:13 +0000 (+0100) Subject: add some folders; more complete package structure X-Git-Url: https://git.auder.net/?p=valse.git;a=commitdiff_plain;h=5ce95f263665997e5319422d19ac2ad9635b1e58;hp=31444abc970b7fe17463bcc916e95846272158db add some folders; more complete package structure --- diff --git a/README.md b/README.md index ada56d8..30d87ce 100644 --- a/README.md +++ b/README.md @@ -5,3 +5,8 @@ This code is the applied part of the PhD thesis of [Benjamin Gohehry](http://www ## Description TODO : see R package + +Trouver un jeu de données (+) intéressant (que les autres) ? +Ajouter toy datasets pour les tests (ou piocher dans MASS ?) + +ED : j'ai simulé un truc basique via 'generateXYdefault(10,5,6,2)' diff --git a/pkg/DESCRIPTION b/pkg/DESCRIPTION index 9d8a677..5a8bc18 100644 --- a/pkg/DESCRIPTION +++ b/pkg/DESCRIPTION @@ -1,17 +1,19 @@ Package: valse -Title: VAriabLe SElection with mixture of models +Title: Variable Selection With Mixture Of Models Date: 2016-12-01 Version: 0.1-0 Description: Two methods are implemented to cluster data with finite mixture regression models. - Those procedures deal with high-dimensional covariates and responses through a variable selection - procedure based on the Lasso estimator. A low-rank constraint could be added, computed for the Lasso-Rank procedure. - A collection of models is constructed, varying the level of sparsity and the number of clusters, and a model is selected - using a model selection criterion (slope heuristic, BIC or AIC). - Details of the procedure are provided in 'Model-based clustering for high-dimensional data. Application to functional data' - by Emilie Devijver, published in Advances in Data Analysis and Clustering (2016) + Those procedures deal with high-dimensional covariates and responses through a variable + selection procedure based on the Lasso estimator. A low-rank constraint could be added, + computed for the Lasso-Rank procedure. + A collection of models is constructed, varying the level of sparsity and the number of + clusters, and a model is selected using a model selection criterion (slope heuristic, + BIC or AIC). Details of the procedure are provided in 'Model-based clustering for + high-dimensional data. Application to functional data' by Emilie Devijver, published in + Advances in Data Analysis and Clustering (2016). Author: Benjamin Auder [aut,cre], + Emilie Devijver [aut], Benjamin Goehry [aut] - Emilie Devijver [aut] Maintainer: Benjamin Auder Depends: R (>= 3.0.0) @@ -20,8 +22,9 @@ Imports: methods Suggests: parallel, - testthat, - knitr + testhat, + devtools, + rmarkdown URL: http://git.auder.net/?p=valse.git License: MIT + file LICENSE VignetteBuilder: knitr diff --git a/pkg/LICENSE b/pkg/LICENSE index 727af26..a212458 100644 --- a/pkg/LICENSE +++ b/pkg/LICENSE @@ -1,6 +1,6 @@ Copyright (c) - 2014-2016, Emilie Devijver 2014-2017, Benjamin Auder + 2014-2017, Emilie Devijver 2016-2017, Benjamin Goehry Permission is hereby granted, free of charge, to any person obtaining diff --git a/pkg/R/plot.R b/pkg/R/plot.R new file mode 100644 index 0000000..a8da583 --- /dev/null +++ b/pkg/R/plot.R @@ -0,0 +1 @@ +#TODO: reprendre les plots d'Emilie dans reports/... diff --git a/pkg/data/TODO b/pkg/data/TODO deleted file mode 100644 index 7e3c7ec..0000000 --- a/pkg/data/TODO +++ /dev/null @@ -1,4 +0,0 @@ -Trouver un jeu de données (+) intéressant (que les autres) ? -Ajouter toy datasets pour les tests (ou piocher dans MASS ?) - -ED : j'ai simulé un truc basique via 'generateXYdefault(10,5,6,2)' diff --git a/pkg/inst/testdata/TODO.csv b/pkg/inst/testdata/TODO.csv new file mode 100644 index 0000000..d679966 --- /dev/null +++ b/pkg/inst/testdata/TODO.csv @@ -0,0 +1 @@ +ou alors data_test.RData, possible aussi diff --git a/pkg/tests/testthat.R b/pkg/tests/testthat.R new file mode 100644 index 0000000..d2761ea --- /dev/null +++ b/pkg/tests/testthat.R @@ -0,0 +1,4 @@ +library(testthat) +library(valse #ou load_all() + +test_check("valse") diff --git a/pkg/tests/testthat/helper-clustering.R b/pkg/tests/testthat/helper-clustering.R new file mode 100644 index 0000000..785b7f0 --- /dev/null +++ b/pkg/tests/testthat/helper-clustering.R @@ -0,0 +1,11 @@ +# Compute the sum of (normalized) sum of squares of closest distances to a medoid. +computeDistortion <- function(series, medoids) +{ + n <- ncol(series) + L <- nrow(series) + distortion <- 0. + for (i in seq_len(n)) + distortion <- distortion + min( colSums( sweep(medoids,1,series[,i],'-')^2 ) / L ) + + sqrt( distortion / n ) +} diff --git a/pkg/tests/testthat/test-clustering.R b/pkg/tests/testthat/test-clustering.R new file mode 100644 index 0000000..2e3a431 --- /dev/null +++ b/pkg/tests/testthat/test-clustering.R @@ -0,0 +1,72 @@ +context("clustering") + +test_that("clusteringTask1 behave as expected", +{ + # Generate 60 reference sinusoïdal series (medoids to be found), + # and sample 900 series around them (add a small noise) + n <- 900 + x <- seq(0,9.5,0.1) + L <- length(x) #96 1/4h + K1 <- 60 + s <- lapply( seq_len(K1), function(i) x^(1+i/30)*cos(x+i) ) + series <- matrix(nrow=L, ncol=n) + for (i in seq_len(n)) + series[,i] <- s[[I(i,K1)]] + rnorm(L,sd=0.01) + + getSeries <- function(indices) { + indices <- indices[indices <= n] + if (length(indices)>0) as.matrix(series[,indices]) else NULL + } + + wf <- "haar" + ctype <- "absolute" + getContribs <- function(indices) curvesToContribs(as.matrix(series[,indices]),wf,ctype) + + require("cluster", quietly=TRUE) + algoClust1 <- function(contribs,K) cluster::pam(t(contribs),K,diss=FALSE)$id.med + indices1 <- clusteringTask1(1:n, getContribs, K1, algoClust1, 140, verbose=TRUE, parll=FALSE) + medoids_K1 <- getSeries(indices1) + + expect_equal(dim(medoids_K1), c(L,K1)) + # Not easy to evaluate result: at least we expect it to be better than random selection of + # medoids within initial series + distor_good <- computeDistortion(series, medoids_K1) + for (i in 1:3) + expect_lte( distor_good, computeDistortion(series,series[,sample(1:n, K1)]) ) +}) + +test_that("clusteringTask2 behave as expected", +{ + # Same 60 reference sinusoïdal series than in clusteringTask1 test, + # but this time we consider them as medoids - skipping stage 1 + # Here also we sample 900 series around the 60 "medoids" + n <- 900 + x <- seq(0,9.5,0.1) + L <- length(x) #96 1/4h + K1 <- 60 + K2 <- 3 + #for (i in 1:60) {plot(x^(1+i/30)*cos(x+i),type="l",col=i,ylim=c(-50,50)); par(new=TRUE)} + s <- lapply( seq_len(K1), function(i) x^(1+i/30)*cos(x+i) ) + series <- matrix(nrow=L, ncol=n) + for (i in seq_len(n)) + series[,i] <- s[[I(i,K1)]] + rnorm(L,sd=0.01) + + getSeries <- function(indices) { + indices <- indices[indices <= n] + if (length(indices)>0) as.matrix(series[,indices]) else NULL + } + + # Perfect situation: all medoids "after stage 1" are ~good + algoClust2 <- function(dists,K) cluster::pam(dists,K,diss=TRUE)$id.med + indices2 <- clusteringTask2(1:K1, getSeries, K2, algoClust2, 210, 3, 4, 8, "little", + verbose=TRUE, parll=FALSE) + medoids_K2 <- getSeries(indices2) + + expect_equal(dim(medoids_K2), c(L,K2)) + # Not easy to evaluate result: at least we expect it to be better than random selection of + # synchrones within 1...K1 (from where distances computations + clustering was run) + distor_good <- computeDistortion(series, medoids_K2) +#TODO: This fails; why? +# for (i in 1:3) +# expect_lte( distor_good, computeDistortion(series, series[,sample(1:K1,3)]) ) +}) diff --git a/pkg/vignettes/valse.Rmd b/pkg/vignettes/valse.Rmd new file mode 100644 index 0000000..e8164a1 --- /dev/null +++ b/pkg/vignettes/valse.Rmd @@ -0,0 +1,23 @@ +--- +title: "valse package vignette" +output: html_document +--- + +```{r include = FALSE} +library(valse) +``` + +The code below demonstrates blabla... in [valse](https://github.com/blabla/valse) package. +Each plot displays blabla... + +## Des jolis plot 1 + +```{r} +#plotBla1(...) +``` + +## Des jolies couleurs 2 + +```{r} +#plotBla2(...) +``` diff --git a/src/adapters/a.EMGLLF.o b/src/adapters/a.EMGLLF.o deleted file mode 100644 index 70d7499..0000000 Binary files a/src/adapters/a.EMGLLF.o and /dev/null differ diff --git a/src/adapters/a.EMGrank.o b/src/adapters/a.EMGrank.o deleted file mode 100644 index d9f1d09..0000000 Binary files a/src/adapters/a.EMGrank.o and /dev/null differ diff --git a/src/adapters/a.constructionModelesLassoMLE.o b/src/adapters/a.constructionModelesLassoMLE.o deleted file mode 100644 index 5c98e9f..0000000 Binary files a/src/adapters/a.constructionModelesLassoMLE.o and /dev/null differ diff --git a/src/adapters/a.constructionModelesLassoRank.o b/src/adapters/a.constructionModelesLassoRank.o deleted file mode 100644 index 539977a..0000000 Binary files a/src/adapters/a.constructionModelesLassoRank.o and /dev/null differ diff --git a/src/adapters/a.selectiontotale.o b/src/adapters/a.selectiontotale.o deleted file mode 100644 index 13adb60..0000000 Binary files a/src/adapters/a.selectiontotale.o and /dev/null differ diff --git a/src/sources/EMGLLF.o b/src/sources/EMGLLF.o deleted file mode 100644 index e50edc6..0000000 Binary files a/src/sources/EMGLLF.o and /dev/null differ diff --git a/src/sources/EMGrank.o b/src/sources/EMGrank.o deleted file mode 100644 index 97a0d30..0000000 Binary files a/src/sources/EMGrank.o and /dev/null differ diff --git a/src/sources/constructionModelesLassoMLE.o b/src/sources/constructionModelesLassoMLE.o deleted file mode 100644 index 38ebfdd..0000000 Binary files a/src/sources/constructionModelesLassoMLE.o and /dev/null differ diff --git a/src/sources/constructionModelesLassoRank.o b/src/sources/constructionModelesLassoRank.o deleted file mode 100644 index dce02cf..0000000 Binary files a/src/sources/constructionModelesLassoRank.o and /dev/null differ diff --git a/src/sources/selectiontotale.o b/src/sources/selectiontotale.o deleted file mode 100644 index 547475d..0000000 Binary files a/src/sources/selectiontotale.o and /dev/null differ diff --git a/src/valse.so b/src/valse.so deleted file mode 100755 index ee29721..0000000 Binary files a/src/valse.so and /dev/null differ