A variable selection method for simultaneous component based data integration

Gu, Zhengguo; Deun, Katrijn Van

doi:10.1016/j.chemolab.2016.07.013

Cited by 12 publications

(24 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The method introduced here builds further on extensions of principal component analysis. These include sparse PCA (Zou et al, 2006), simultaneous components with rotation to common and distinctive components (Schouteden et al, 2013), and sparse simultaneous component analysis (Gu & Van Deun, 2016; Van Deun, Wilderjans, van den Berg, Antoniadis, & Van Mechelen, 2011).…”

Section: Methodsmentioning

confidence: 99%

See 1 more Smart Citation

Revealing the Joint Mechanisms in Traditional Data Linked With Big Data

Schipper

Deun

2018

Zeitschrift für Psychologie

Self Cite

View full text Add to dashboard Cite

Abstract. Recent technological advances have made it possible to study human behavior by linking novel types of data to more traditional types of psychological data, for example, linking psychological questionnaire data with genetic risk scores. Revealing the variables that are linked throughout these traditional and novel types of data gives crucial insight into the complex interplay between the multiple factors that determine human behavior, for example, the concerted action of genes and environment in the emergence of depression. Little or no theory is available on the link between such traditional and novel types of data, the latter usually consisting of a huge number of variables. The challenge is to select – in an automated way – those variables that are linked throughout the different blocks, and this eludes currently available methods for data analysis. To fill the methodological gap, we here present a novel data integration method.

show abstract

Section: Methodsmentioning

confidence: 99%

“…Recently, Gu and Van Deun (2016) developed an extension to sparse SCA by penalizing the loading matrix in a componentwise fashion, hence allowing for both common and distinctive components. The main distinguishing characteristic of this paper is that it penalizes the component weights and not the loadings.…”

Section: Methodsmentioning

confidence: 99%

Revealing the Joint Mechanisms in Traditional Data Linked With Big Data

Schipper

Deun

2018

Zeitschrift für Psychologie

Self Cite

View full text Add to dashboard Cite

show abstract

“…Selecting the proper parameters in a non-simulation study, especially when analysing high-dimensional data, is challenging. As demonstrated by (Gu & Van Deun, 2016), the tuning parameter of l1 -Lasso can successfully be selecting using (Meinshausen & Bühlmann, 2010)'s resample-based stability selection method, although the multi-component structure of both (Gu & Van Deun, 2016)'s and our work complicates matters further.…”

Section: Selection Status Recoverymentioning

confidence: 96%

“…To single out the cross-source relations, we need to disentangle the common sources of variation shared between the different data blocks (with each block containing the data or variables of one source) from the sources of variation that are specific for a single or a few data blocks only. To do this, we perform a so called sparse DIStinctive and COmmon Simultaneus Component Analysis decomposition of the data (sparse DISCO SCA; see (Gu & Van Deun, 2016, 2017), a method that was developed for the integrated analysis of multi-source data with the specific aim of separating block-specific sources of variation from common sources of variation. Sparse DISCO SCA models common and specific components by using specific constraints on the loadings of each of the components in Λ.…”

Section: Isolating the Cross-source Relationsmentioning

confidence: 99%

Introducing SNAC: Sparse Network and Component model for integration of multi-source data

Tio¹,

Waldorp

Deun³

2018

Preprint

Self Cite

View full text Add to dashboard Cite

Gaussian graphical models (GGMs) are a popular method for analysing complex data by modelling the unique relationships between variables. Recently, a shift in interest has taken place from investigating relationships within a discipline (e.g. genetics) to estimating relationships between variables from various disciplines (e.g. how gene expression relates to cognitive performance). It is thus not surprising that there is an increasing need for analysing large, so-called \textit{multi-source} datasets, each containing detailed information from many data sources on the same individuals. GGMs are a straightforward statistical candidate for estimating \textit{unique cross-source relationships} from such network-oriented data. However, the multi-source nature of the data poses two challenges: First, different sources may inherently differ from one another, biasing the estimated relations. Second, GGMs are not cut out for separating cross-source relationships from all other, source-specific relationships. In this paper we propose adding a simultaneous-component-model as a pre-pocessing step to the GGM, the combination of which is suitable for estimating cross-source relationships from multi-source data. Compared to the graphical lasso (a commonly used GGM technique), this Sparse Network And Component (SNAC) model more accurately estimates the unique cross-source relationships from multi-source data. This holds in particular when the data contains more variables than observations ($p>n$). Neither differences in sparseness of the underlying component structure of the data nor in the relative dominance of the cross-source compared to source-specific relationships strongly affect the relationship estimates. Sparse Network And Component analysis, a hybrid component-graphical model, is a promising tool for modelling unique relationships between different data sources, thus providing insight in how various disciplines are connected to one another.

show abstract

“…This article introduces an R package for performing joint analysis on large-scale multiblock data from multiple sources. The core algorithms of this package have their roots in traditional simultaneous component analysis (SCA), which has been widely used for performing data integration from multiple sources in biomedical research, bioinformatics, genomics, and psychology (e.g., De Tayrac, Lê, Aubry, Mosser, & Husson, 2009; Gu & Van Deun, 2016; Lock, Hoadley, Marron, & Nobel, 2013; Van Deun, Smilde, van der Werf, Kiers, & Van Mechelen, 2009; Van Deun et al, 2012: Van Deun, Smilde, Thorrez, Kiers, & Van Mechelen, 2013; Wilderjans, Ceulemans, Van Mechelen, & van den Berg, 2011). One may notice that, aside from SCA, other methods, such as canonical correlation analysis (Tenenhaus & Tenenhaus, 2014), may also be used for joint analysis of multiblock data, but we refrain from discussing other methods in this article.…”

mentioning

confidence: 99%

RegularizedSCA: Regularized simultaneous component analysis of multiblock data in R

Deun

2018

Behav Res

Self Cite

View full text Add to dashboard Cite

This article introduces a package developed for R (R Core Team, 2017) for performing an integrated analysis of multiple data blocks (i.e., linked data) coming from different sources. The methods in this package combine simultaneous component analysis (SCA) with structured selection of variables. The key feature of this package is that it allows to (1) identify joint variation that is shared across all the data sources and specific variation that is associated with one or a few of the data sources and (2) flexibly estimate component matrices with predefined structures. Linked data occur in many disciplines (e.g., biomedical research, bioinformatics, chemometrics, finance, genomics, psychology, and sociology) and especially in multidisciplinary research. Hence, we expect our package to be useful in various fields.

show abstract

A variable selection method for simultaneous component based data integration

Cited by 12 publications

References 23 publications

Revealing the Joint Mechanisms in Traditional Data Linked With Big Data

Revealing the Joint Mechanisms in Traditional Data Linked With Big Data

Introducing SNAC: Sparse Network and Component model for integration of multi-source data

RegularizedSCA: Regularized simultaneous component analysis of multiblock data in R

Contact Info

Product

Resources

About