ggRandomForests: Visually Exploring Random Forests

ggRandomForests will help uncover variable associations in the random forests models. The package is designed for use with the randomForest package (A. Liaw and M. Wiener 2002) or the randomForestSRC package (Ishwaran et.al. 2014, 2008, 2007) for survival, regression and classification random forests and uses the ggplot2 package (Wickham 2009) for plotting diagnostic and variable association results. ggRandomForests is structured to extract data objects from randomForestSRC or randomForest objects and provides S3 functions for printing and plotting these objects.

The randomForestSRC package provides a unified treatment of Breiman's (2001) random forests for a variety of data settings. Regression and classification forests are grown when the response is numeric or categorical (factor) while survival and competing risk forests (Ishwaran et al. 2008, 2012) are grown for right-censored survival data. Recently, support for the randomForest package (A. Liaw and M. Wiener 2002) for regression and classification forests has also been added.

Many of the figures created by the ggRandomForests package are also available directly from within the randomForestSRC or randomForest package. However, ggRandomForests offers the following advantages:

• Separation of data and figures: ggRandomForests contains functions that operate on either the forest object directly, or on the output from randomForestSRC and randomForest post processing functions (i.e. plot.variable, var.select, find.interaction) to generate intermediate ggRandomForests data objects. S3 functions are provide to further process these objects and plot results using the ggplot2 graphics package. Alternatively, users can use these data objects for additional custom plotting or analysis operations.

• Each data object/figure is a single, self contained object. This allows simple modification and manipulation of the data or ggplot2 objects to meet users specific needs and requirements.

• The use of ggplot2 for plotting. We chose to use the ggplot2 package for our figures to allow users flexibility in modifying the figures to their liking. Each S3 plot function returns either a single ggplot2 object, or a list of ggplot2 objects, allowing users to use additional ggplot2 functions or themes to modify and customize the figures to their liking.

The package has recently been extended for Breiman and Cutler's Random Forests for Classification and Regression package randomForest where possible. Though methods have been provided for all gg_* functions, the unsupported functions will return an error message indicating where support is still lacking.

References

Breiman, L. (2001). Random forests, Machine Learning, 45:5-32.

Ishwaran H. and Kogalur U.B. (2014). Random Forests for Survival, Regression and Classification (RF-SRC), R package version 1.5.5.

Ishwaran H. and Kogalur U.B. (2007). Random survival forests for R. R News 7(2), 25--31.

Ishwaran H., Kogalur U.B., Blackstone E.H. and Lauer M.S. (2008). Random survival forests. Ann. Appl. Statist. 2(3), 841--860.

A. Liaw and M. Wiener (2002). Classification and Regression by randomForest. R News 2(3), 18--22.

Wickham, H. ggplot2: elegant graphics for data analysis. Springer New York, 2009.