Consistent selection of the number of clusters via crossvalidation

Wang, Junhui

doi:10.1093/biomet/asq061

Cited by 145 publications

(127 citation statements)

References 16 publications

Supporting

Mentioning

124

Contrasting

Order By: Relevance

“…In most clustering algorithms, however, the decision on the number of clusters is a difficult task. Its choice as a stochastic parameter usually complicates modeling and increases the computational burden [15,16,17].…”

Section: Introductionmentioning

confidence: 99%

Bayesian nonparametric clustering and association studies for candidate SNP observations

Wang

Ruggeri

Hsiao

et al. 2017

International Journal of Approximate Reasoning

View full text Add to dashboard Cite

Clustering is often considered as the first step in the analysis when dealing with an enormous amount of Single Nucleotide Polymorphism (SNP) genotype data. The lack of biological information could affect the outcome of such procedure. Even if a clustering procedure has been selected and performed, the impact of its uncertainty on the subsequent association analysis is rarely assessed. In this research we propose first a model to cluster SNPs data, then we assess the association between the cluster and a disease. In particular, we adopt a Dirichlet process mixture model with the advantages, with respect to the usual clustering methods, that the number of clusters needs not to be known and fixed in advance and the variation in the assignment of SNPs to clusters can be accounted. In addition, once a clustering of SNPs is obtained, we design an individualized genetic score quantifying the SNP composition in each cluster for every subject, so that we can set up a generalized linear model for association analysis able to incorporate the information from a large-scale SNP dataset, and yet with a much smaller number of explanatory variables. The inference on cluster allocation, the strength of association of each cluster (the collective effect on SNPs in the same cluster), and the susceptibility of each SNP are based on posterior samples from Markov chain Monte Carlo methods and the Binder loss information. We exemplify this Bayesian nonparametric strategy in a genome-wide association study of Crohn's * Corresponding author.Email address: raffaele@mi.imati.cnr.it (Raffaele Argiento)Preprint submitted to International Journal of Approximate Reasoning July 16, 2016 disease in a case-control setting.

show abstract

Section: Introductionmentioning

confidence: 99%

Bayesian nonparametric clustering and association studies for candidate SNP observations

Wang

Ruggeri

Hsiao

et al. 2017

International Journal of Approximate Reasoning

View full text Add to dashboard Cite

show abstract

“…Consensus Clustering [76] and Resampling [77] try to find k looking for the most "stable" configuration through different MonteCarlo simulations but with the same number of clusters. On the contrary, Wang [78] proposed selecting the number of clusters by minimizing the algorithm's instability, a simple measure of the robustness of any algorithm against the initial random seeds.…”

Section: Clustering Algorithmsmentioning

confidence: 99%

Explorative Multidimensional Analysis for Energy Efficiency: DataViz versus Clustering Algorithms

Cottafava

Sonetti²,

Gambino

et al. 2018

Energies

View full text Add to dashboard Cite

Abstract:We propose a simple tool to help the energy management of a large building stock defining clusters of buildings with the same function, setting alert thresholds for each cluster, and easily recognizing outliers. The objective is to enable a building management system to be used for detection of abnormal energy use. We start reviewing energy performance indicators, and how they feed into data visualization (DataViz) tools for a large building stock, especially for university campuses. After a brief presentation of the University of Turin's building stock which represents our case study, we perform an explorative analysis based on the Multidimensional Detective approach by Inselberg, using the Scatter Plot Matrix and the Parallel Coordinates methods. The k-means clustering algorithm is then applied on the same dataset to test the hypotheses made during the explorative analysis. Our results show that DataViz techniques provide quick and user-friendly solutions for the energy management of a large stock of buildings. In particular, they help identifying clusters of buildings and outliers and setting alert thresholds for various Energy Efficiency Indices.

show abstract

“…The authors focus on the concept of stability as robustness to randomness present in the sample. Drawing on the work of Wang (2010), they formulate the concept of stability in the following way: if one draws samples from the population and applies a selected clustering algorithm, the results of grouping should not be very different.…”

Section: Package Fpcmentioning

confidence: 99%

Using R Packages for Comparison of Cluster Stability

Rozmus

2017

Folia Oeconomica

View full text Add to dashboard Cite

The stability of clustering methods is the issue that has attracted a considerable amount of attention of researchers in recent years. In this respect, the major question that needs to be answered seems to be to what extent the structure discovered by a particular method is actually present in the data. The literature proposes a number of different ways of measuring stability. The theoretical considerations have led to the development of computer tools for the practical implementation of the proposed ways to study stability. The practical tools are available within several R packages, for example, clv, clValid, fpc, ClusterStability, and pvclust. Due to the hypothesis that cluster stability can be the answer to the question about the right number of groups in clustering, the main aim of this article is to compare the results of the studies on clustering stability conducted with three R packages, i.e.: clv, clValid, and fpc.

show abstract

Consistent selection of the number of clusters via crossvalidation

Cited by 145 publications

References 16 publications

Bayesian nonparametric clustering and association studies for candidate SNP observations

Bayesian nonparametric clustering and association studies for candidate SNP observations

Explorative Multidimensional Analysis for Energy Efficiency: DataViz versus Clustering Algorithms

Using R Packages for Comparison of Cluster Stability

Contact Info

Product

Resources

About