ggRandomForests (version 1.0.0)

ggRandomForests: ggRandomForests - package for plotting of randomForestSRC objects using the ggplot2 package.

Description

ggRandomForests is a utility package for randomForestSRC. The package is designed to simplify the graphical analysis and exploration of randomForests.

The randomForestSRC package provides a unified treatment of Breimans random forests (Breiman 2001) 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. Support for unsupervised and multivariate randomForests have also recently been added.

Many of the features of the ggRandomForests package are available within the randomForestSRC package. However, the ggRandomForests offers the following advantages:

• Separation of data and figures: ggRandomForest contains functions that operate on either the raw randomForestSRC (rfsrc) forest object directly, or on the output from rfsrc posprocessing functions (i.e.vimp.rfsrc,var.select.rfsrc,plot.variable.rfsrc) to generate intermediate data.frame objects. These objects are then passed to corresponding plot functions using the S3 object model. Alternatively, a user can use these data object for additional external, custom plotting or analysis operations.
• Each figure is self contained. This allows modification of individual plots to meet the users specification. Figures can be grouped together using functions from thegridExtrapackage.
• ggplot2figures: We chose to use theggplot2package for our figures. The plot functions all return either a singleggplot2object, or alistofggplot2objects. The user can then use additionalggplot2functions to modify and customise the figures to their liking.

The ggRandomForests package contains the following functions:

• gg_rfsrc: randomForest[SRC] prediction
• gg_error: randomForest[SRC] convergence rate based on the OOB error rate.
• gg_roc: ROC curves for randomForest classification models.
• gg_vimp: Variable Importance ranking for variable selection
• gg_minimal_depth: Minimal Depth ranking for variable selection
• gg_interaction: Minimal Depth interaction detection (under development)
• gg_variable: Marginal variable dependence (including conditional dependence)
• gg_partial: Partial variable dependence (including conditional partial dependence)
• gg_survival: Random Forest survival plots including either Kaplan-Meier or Nelson-Aalon survival estimates

All functions have an associated plotting function that returns ggplot2 graphics, either individually or as a list, that can be further customised using standard ggplot2 commands.

References

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

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

Ishwaran, H. and Kogalur, U.B. (2014). Random Forests for Survival, Regression and Classification (RF-SRC), R package version 1.5.

Ishwaran, H., U. B. Kogalur, E. Z. Gorodeski, A. J. Minn, and M. S. Lauer (2010). High-dimensional variable selection for survival data. J. Amer. Statist. Assoc. 105, 205-217.

Ishwaran, H. (2007). Variable importance in binary regression trees and forests. Electronic J. Statist., 1, 519-537.

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