Title: | VALID Inference for Clusters Separation Testing |
Version: | 0.1.0 |
Author: | Benjamin Hivert |
Maintainer: | Benjamin Hivert <benjamin.hivert@u-bordeaux.fr> |
Description: | Given a partition resulting from any clustering algorithm, the implemented tests allow valid post-clustering inference by testing if a given variable significantly separates two of the estimated clusters. Methods are detailed in: Hivert B, Agniel D, Thiebaut R & Hejblum BP (2022). "Post-clustering difference testing: valid inference and practical considerations", <doi:10.48550/arXiv.2210.13172>. |
Encoding: | UTF-8 |
RoxygenNote: | 7.2.2 |
Imports: | diptest, dplyr |
Depends: | R (≥ 3.6) |
License: | MIT + file LICENSE |
NeedsCompilation: | no |
Packaged: | 2022-11-30 15:06:52 UTC; benjaminhivert |
Repository: | CRAN |
Date/Publication: | 2022-12-01 08:20:02 UTC |
Merged version of the selective test
Description
Merged version of the selective test
Usage
merge_selective_inference(X, k1, k2, g, ndraws = 2000, cl_fun, cl)
Arguments
X |
The data matrix of size on which the clustering is applied |
k1 |
The first cluster of interest |
k2 |
The second cluster of interest |
g |
The variables for which the test is applied |
ndraws |
The number of Monte-Carlo samples |
cl_fun |
The clustering function used to build clusters |
cl |
The labels of the data obtained thanks to the |
Value
A list with the following elements
-
pval
: The resulting p-values of the test. -
adjacent
: List of the adjacent clusters between k1 and k2 -
pval_adj
: The corresponding adjacent p-values that are merged
Examples
X <- matrix(rnorm(200),ncol = 2)
hcl_fun <- function(x){
return(as.factor(cutree(hclust(dist(x), method = "ward.D2"), k=4)))}
cl <- hcl_fun(X)
plot(X, col=cl)
#Note that in practice the value of ndraws (the number of Monte-Carlo simulations must be higher)
test_var1 <- test_selective_inference(X, k1=1, k2=4, g=1, ndraws =100, cl_fun = hcl_fun, cl = cl)
Multimodality test for post clustering variable involvement
Description
Multimodality test for post clustering variable involvement
Usage
test_multimod(X, g, cl, k1, k2)
Arguments
X |
The data matrix of size on which the clustering is applied |
g |
The variable on which the test is applied |
cl |
The labels of the data obtained thanks to a clustering algorithm |
k1 |
The first cluster of interest |
k2 |
The second cluster of interest |
Value
A list containing : A list with the following elements
-
data_for_test
: The data used for the test -
stat_g
: The dip statistic -
pval
: The resulting p-values of the test computed with thediptest
function
Examples
X <- matrix(rnorm(200),ncol = 2)
hcl_fun <- function(x){
return(as.factor(cutree(hclust(dist(x), method = "ward.D2"), k=2)))}
cl <- hcl_fun(X)
plot(X, col=cl)
test_var1 <- test_multimod(X, g=1, k1=1, k2=2, cl = cl)
test_var2 <- test_multimod(X, g=2, k1=1, k2=2, cl = cl)
Selective inference for post-clustering variable involvement
Description
Selective inference for post-clustering variable involvement
Usage
test_selective_inference(
X,
k1,
k2,
g,
ndraws = 2000,
cl_fun,
cl = NULL,
sig = NULL
)
Arguments
X |
The data matrix of size on which the clustering is applied |
k1 |
The first cluster of interest |
k2 |
The second cluster of interest |
g |
The variables for which the test is applied |
ndraws |
The number of Monte-Carlo samples |
cl_fun |
The clustering function used to build clusters |
cl |
The labels of the data obtained thanks to the |
sig |
The estimated standard deviation. Default is NULL and the standard deviation is estimated using only observations in the two clusters of interest |
Value
A list with the following elements
-
stat_g
: the test statistic used for the test. -
pval
: The resulting p-values of the test. -
stder
: The standard deviation of the p-values computed thanks to the Monte-Carlo samples. -
clusters
: The labels of the data.
Note
This function is adapted from the clusterpval::test_clusters_approx() of Gao et al. (2022) (available on Github: https://github.com/lucylgao/clusterpval)
References
Gao, L. L., Bien, J., & Witten, D. (2022). Selective inference for hierarchical clustering. Journal of the American Statistical Association, (just-accepted), 1-27.
Examples
X <- matrix(rnorm(200),ncol = 2)
hcl_fun <- function(x){
return(as.factor(cutree(hclust(dist(x), method = "ward.D2"), k=2)))}
cl <- hcl_fun(X)
plot(X, col=cl)
#Note that in practice the value of ndraws (the number of Monte-Carlo simulations must be higher)
test_var1 <- test_selective_inference(X, k1=1, k2=2, g=1, ndraws =100, cl_fun = hcl_fun, cl = cl)