Multiple Additive Regression Trees




MART(tm) is an implementation of the gradient tree boosting methods for predictive data mining (regression and classification) described in Greedy Function Approximation: a Gradient Boosting Machine (PostScript), and Stochastic Gradient Boosting (PostScript).

An interface for using MART(tm) with Splus2000(r) [Statistical Sciences, Inc., Seattle, WA]  can be obtained from Salford Systems. It runs on PC compatible computers with a Microsoft operating system (Windows 95/98/NT/2000/XT), or on Silicon Graphics (SGI) and Sun Microsystems (SUN) computers running UNIX.


A tutorial introduction Tutorial: Getting Started with MART in Splus (PostScript) illustrates the basic facilities of the MART/Splus interface and how to use them. It presumes minimal knowledge concerning the technical details of the methodology or the use of Splus. The following data files are used in the tutorial: xdata, ydata, classlab. These should be saved in *.txt format.


Complete documentation for each MART/Splus procedure is provided with associated help files, as described in the readme.txt file provided with the installation. Some familiarity with the above two papers is helpful when experimenting with some of the more advanced features and options.


Coming soon: Linux implementation.