README.md

In gastonstat/blockberry: Wireframe for multiblock objects

blockberry

blockberry is an experimental R package that provides a wireframe for working with multiblock objects.

Motivation

The main motivating question behind blockberry is: how to handle multiblock data?

Our approach for addressing the previous question is based on an extremely simple yet ingenuous concept: take into account the block structure by means of what we call block-dimension

Simply put, the block-dimension is implemented as an attribute in R objects that allows us to introduce a block or partitioned structure.

Installation

Since blockberry is an experimental in-progress package, its distribution is not on CRAN but on the github repository https://github.com/gastonstat/blockberry

In order to install blockberry you need to use the function install_github() from the package devtools (remember to install it first). Type the following lines in your R console:

# only if you haven't installed "devtools"
install.packages("devtools")

# load "devtools"
library(devtools)

# install "blockberry"
install_github('blockberry', username = 'gastonstat')

# load "blockberry
library(blockberry)

Example with blockvector

How to create a blockvector

# say you have a numeric vector
vnum = 1:10

# blockvector (partitioned in 3 blocks)
bnum = blockvector(vnum, parts = c(3, 2, 5), dims = 3)

Example with blockmatrix

How to create a blockmatrix

# say you have a numeric matrix
m = matrix(1:20, 4, 5)

# option 1) blockmatrix using arguments `rowparts` and `colparts`
bm1 = blockmatrix(m, rowparts = c(2, 2), colparts = c(3, 2))

# option 2) blockmatrix using arguments `parts` and `dims`
bm2 = blockmatrix(vnum, parts = c(2, 2, 3, 2), dims = c(2, 2))

Authors Contact

Gaston Sanchez (gaston.stat at gmail.com)

Mohamed Hanafi (mohamed.hanafi at oniris-nantes.fr)

gastonstat/blockberry documentation built on May 16, 2019, 5:44 p.m.