checkClusterability {DAPAR} | R Documentation |
The first step is to standardize the data (with the Mfuzz
package). Then the function checks that these data are clusterizable or not
(use of [diptest::dip.test()] to determine whether the distribution is unimodal or
multimodal). Finally, it determines the "optimal" k by the Gap statistic
approach.
checkClusterability(standards, b = 500)
standards |
a matrix or dataframe containing only the standardized mean intensities returned by the function [standardiseMeanIntensities()] |
b |
Parameter B of the function [gap_cluster()] |
a list of 2 elements: * dip_test: the result of the clusterability of the data * gap_cluster: the gap statistic obtained with the function [cluster::clusGap()].
Helene Borges
utils::data(Exp1_R25_prot, package='DAPARdata') obj <- Exp1_R25_prot[1:1000] keepThat <- mvFilterGetIndices(obj, condition = 'WholeMatrix', threshold=ncol(obj)) obj <- mvFilterFromIndices(obj, keepThat) averaged_means <- averageIntensities(obj) only_means <- dplyr::select_if(averaged_means, is.numeric) only_features <- dplyr::select_if(averaged_means, is.character) means <- purrr::map(purrr::array_branch(as.matrix(only_means), 1),mean) centered <- only_means - unlist(means) centered_means <- dplyr::bind_cols(feature = dplyr::as_tibble(only_features), dplyr::as_tibble(centered)) checkClust <- checkClusterability(centered_means, b=100)