contextual: Simulation and Analysis of Contextual Multi-Armed Bandit Policies

Facilitates the simulation and evaluation of context-free and contextual multi-Armed Bandit policies or algorithms to ease the implementation, evaluation, and dissemination of both existing and new bandit algorithms and policies.

Depends: R (≥ 3.5.0)
Imports: R6 (≥ 2.3.0), data.table, R.devices, foreach, doParallel, itertools, iterators, Formula, rjson
Suggests: testthat, RCurl, splitstackshape, covr, knitr, here, rmarkdown, devtools, ggplot2, vdiffr
Published: 2020-07-25
Author: Robin van Emden ORCID iD [aut, cre], Maurits Kaptein ORCID iD [ctb]
Maintainer: Robin van Emden <robinvanemden at>
License: GPL-3
NeedsCompilation: no
Materials: README NEWS
CRAN checks: contextual results


Reference manual: contextual.pdf
Vignettes: Demo: Basic Synthetic cMAB Policies
Demo: Offline cMAB LinUCB evaluation
Demo: MAB Replication Eckles & Kaptein (Bootstrap Thompson Sampling)
Demo: Basic Epsilon Greed
Getting started: running simulations
Demo: MAB Policies Comparison
Demo: MovieLens 10M Dataset
Demo: Offline cMAB: CarsKit DePaul Movie Dataset
Offline evaluation: Replication of Li et al 2010
Demo: Bandits, Propensity Weighting & Simpson's Paradox in R
Demo: Replication Sutton & Barto, Reinforcement Learning: An Introduction, Chapter 2
Demo: Replication of John Myles White, Bandit Algorithms for Website Optimization
Package source: contextual_0.9.8.4.tar.gz
Windows binaries: r-devel:, r-release:, r-oldrel:
macOS binaries: r-release: contextual_0.9.8.4.tgz, r-oldrel: contextual_0.9.8.4.tgz
Old sources: contextual archive


Please use the canonical form to link to this page.