CRAN_Status_Badge Research software
impact Travis-CI Build
Status AppVeyor Build
Status R-CMD-check

Implements algorithms for learning discrete Bayesian network classifiers from data, as well as functions for using these classifiers for prediction, assessing their predictive performance, and inspecting and analyzing their properties.


Load a data set and learn a one-dependence estimator by maximizing Akaike’s information criterion (AIC) score.

tn <- tan_cl('class', car, score = 'aic')
#>   Bayesian network classifier (only structure, no parameters)
#>   class variable:        class 
#>   num. features:   6 
#>   num. arcs:   9 
#>   learning algorithm:    tan_cl

After learning the network’s parameters, you can use it to classify data.

tn <- lp(tn, car, smooth = 0.01)
p <- predict(tn, car, prob = TRUE)
#>      unacc          acc         good        vgood
#> [1,]     1 3.963694e-09 5.682130e-09 4.269700e-09
#> [2,]     1 1.752769e-09 3.310473e-12 3.236335e-09
#> [3,]     1 3.730170e-09 1.090296e-08 1.800719e-12
#> [4,]     1 3.963694e-09 5.682130e-09 4.269700e-09
#> [5,]     1 4.562294e-09 6.965323e-09 4.536532e-09
#> [6,]     1 4.281155e-09 5.366306e-09 5.168828e-09
p <- predict(tn, car, prob = FALSE)
#> [1] unacc unacc unacc unacc unacc unacc
#> Levels: unacc acc good vgood

Estimate predictive accuracy with cross validation.

cv(tn, car, k = 10)
#> [1] 0.9386534

Or compute the log-likelihood

logLik(tn, car)
#> 'log Lik.' -13280.39 (df=131)


Make sure you have at least version 3.2.0 of R. You can install bnclassify from CRAN:


Or get the current development version from Github:

# install.packages('devtools')
# devtools::install_github('bmihaljevic/bnclassify', build_vignettes = TRUE)

Ideally, you would use the build_vignettes = TRUE version, and thus get the vignettes, but it requires programs such as texi2dvi to be installed on your side.

For network plotting and prediction with incomplete data you will also need two packages from Bioconductor. Install them with:

biocLite(c("graph", "Rgraphviz"))


See an overview of the package and examples of usage:

vignette('overview', package = 'bnclassify')

See the list of main functionalities.


Use the usage vignette for more details on the functions.

vignette('usage', package = 'bnclassify')

Then have a look at the remaining vignettes.


Try the bnclassify package in your browser

Any scripts or data that you put into this service are public.

bnclassify documentation built on Nov. 16, 2022, 5:08 p.m.