# pcaMethods v1.64.0

## A collection of PCA methods

Provides Bayesian PCA, Probabilistic PCA, Nipals PCA, Inverse Non-Linear PCA and the conventional SVD PCA. A cluster based method for missing value estimation is included for comparison. BPCA, PPCA and NipalsPCA may be used to perform PCA on incomplete data as well as for accurate missing value estimation. A set of methods for printing and plotting the results is also provided. All PCA methods make use of the same data structure (pcaRes) to provide a common interface to the PCA results. Initiated at the Max-Planck Institute for Molecular Plant Physiology, Golm, Germany.

# pcaMethods

R package for performing principal component analysis PCA with applications to missing value imputation. Provides a single interface to performing PCA using

• SVD: a fast method which is also the standard method in R but which is not applicable for data with missing values.
• NIPALS: an iterative fast method which is applicable also to data with missing values.
• PPCA: Probabilistic PCA which is applicable also on data with missing values. Missing value estimation is typically better than NIPALS but also slower to compute and uses more memory. A port to R of the implementation by Jakob Verbeek.
• BPCA: Bayesian PCA which performs very well in the presence of missing values but is slower than PPCA. A port of the matlab implementation by Shigeyuki Oba.
• NLPCA: Non-linear PCA which can find curves in data and in presence of such can perform accurate missing value estimation. Matlab port of the implementation by Mathias Scholz.

pcaMethods is a Bioconductor package and you can install it by

source("https://bioconductor.org/biocLite.R")
biocLite("pcaMethods")

## Documentation

browseVignettes("pcaMethods")
?<function_name>

## Functions in pcaMethods

 Name Description metaboliteDataComplete A complete metabolite data set from an Arabidopsis coldstress experiment nipalsPca NIPALS PCA R2cum,pcaRes-method Cumulative R2 is the total ratio of variance that is being explained by the model helix A helix structured toy data set loadings,ANY-method Crude way to unmask the function with the same name from stats Q2 Cross-validation for PCA deletediagonals Delete diagonals checkData Do some basic checks on a given data matrix asExprSet Convert pcaRes object to an expression set errorHierarchic Later nVar,pcaRes-method Get the number of variables used to build the PCA model. center,pcaRes-method Get the centers of the original variables cvstat,pcaRes-method Get cross-validation statistics (e.g. $Q^2$). simpleEllipse Hotelling's T^2 Ellipse pcaNet Class representation of the NLPCA neural net predict-methods Predict values from PCA. biplot-methods Plot a overlaid scores and loadings plot linr Linear kernel pcaRes Class for representing a PCA result R2VX,pcaRes-method R2 goodness of fit derrorHierarchic Later showNniRes Print a nniRes model nlpca Non-linear PCA pca Perform principal component analysis lineSearch Line search for conjugate gradient llsImpute LLSimpute algorithm scl,pcaRes-method Get the scales (e.g. standard deviations) of the original variables nP,pcaRes-method Get number of PCs DModX,pcaRes-method DModX method,pcaRes-method Get the used PCA method dim.pcaRes Dimensions of a PCA model fitted-methods Extract fitted values from PCA. nPcs,pcaRes-method Get number of PCs. ppca Probabilistic PCA BPCA_dostep Do BPCA estimation step plotPcs Plot many side by side scores XOR loadings plots loadings,pcaRes-method Get loadings from a pcaRes object summary Summary of PCA model sDev,pcaRes-method Get the standard deviations of the scores (indicates their relevance) robustSvd Alternating L1 Singular Value Decomposition show-methods Print/Show for pcaRes leverage,pcaRes-method Extract leverages of a PCA model pcaMethods-deprecated Deprecated methods for pcaMethods weightsAccount Create an object that holds the weights for nlpcaNet. Holds and sets weights in using an environment object. scores.pcaRes Get scores from a pcaRes object vector2matrices,matrix-method Tranform the vectors of weights to matrix structure metaboliteData A incomplete metabolite data set from an Arabidopsis coldstress experiment pcaMethods pcaMethods tempFixNas Temporary fix for missing values optiAlgCgd Conjugate gradient optimization plot.pcaRes Plot diagnostics (screeplot) orth Calculate an orthonormal basis vector2matrices,nlpcaNet-method Tranform the vectors of weights to matrix structure svdImpute SVDimpute algorithm svdPca Perform principal component analysis using singular value decomposition nni Nearest neighbour imputation repmat Replicate and tile an array. scaled,pcaRes-method Check if scaling was part of the PCA model wasna,pcaRes-method Get a matrix with indicating the elements that were missing in the input data. Convenient for estimating imputation performance. centered,pcaRes-method Check centering was part of the model cvseg Get CV segments robustPca PCA implementation based on robustSvd nObs,pcaRes-method Get the number of observations used to build the PCA model. loadings.pcaRes Get loadings from a pcaRes object bpca Bayesian PCA missing value estimation BPCA_initmodel Initialize BPCA model RnipalsPca NIPALS PCA implemented in R kEstimate Estimate best number of Components for missing value estimation kEstimateFast Estimate best number of Components for missing value estimation nmissing,pcaRes-method Missing values nniRes Class for representing a nearest neighbour imputation result sortFeatures Sort the features of NLPCA object prep Pre-process a matrix for PCA scores,pcaRes-method Get scores from a pcaRes object getHierarchicIdx Index in hiearchy forkNlpcaNet Complete copy of nlpca net object completeObs,nniRes-method Get the original data with missing values replaced with predicted values. listPcaMethods List PCA methods slplot,pcaRes-method Side by side scores and loadings plot rediduals-methods Residuals values from a PCA model. No Results!