资 源 简 介
ptucker-em is an R implementation of pTucker (Chu and Ghahramani, 2010) with the EM algorithm. Like PCA or matrix factorization, ptucker-em finds a low-dimensional representation of a given data array (tensor).
Keywords: tensor decomposition, missing-values prediction (imputation), feature extraction, Bayesian probabilistic model.
Installation
Install tensorA package from CRAN. Run R and type
> install.packages("tensorA")
Include the source code pTucker.R.
> source("path/to/pTucker.R")
Usage
Model fitting
pTucker(X, ranks, ...) performs to solve a data array X by a pTucker model in which the dimensions are specified by ranks.
Details of the arguments of pTucker():
* X: An L-dimensional array. "NA" elements are considered as missings. "L"
