Multivariate ADTree is multivariate extensions to Alternating Decision Tree (ADTree). These variants are Fisher's ADTree, Sparse ADTree, and Regularized Logistic ADTree. Regularization techniques such as Ridge, Lasso, and Elastic Net are incorporated into ADTree, a special decision tree induced based on boosting algorithm. Multivariate ADTree is also a Matlab toolbox and it has been developed by Hong Kuan Sok, Melanie Po-Leen Ooi, Ye Chow Kuang, and Serge Demidenko in year 2015. The Multivariate ADTree consists of a series of MATLAB functions. SpaSM toolbox** and Statistics and Machine Learning Toolbox are required to run the program. Here's a brief instruction on how to use the Multivariate ADTree:
Step1: Load the dataset in MATLAB
Step2: Set the parameters
Step3: Draw ADTree tree structure and decision rules are written in the specified file, i.e.,:
Examples are given in the zip file. Please refer to "demoADTree.m" and "demoLADTree.m".
**Sparse Statistical Modelling (SpaSM) is developed by the Department of Informatics at the Technical University of Denmark. The ADTree has no involvement in the development of SpaSM. Please visit the SpaSM link to understand the toolbox.
|CLICK HERE TO DOWNLOAD MULTIVARIATE ADTREE TOOLBOX|
The figures below show the tree diagram constructed using the Breast cancer dataset:
The decision rule of each nodes obtained after the ADTree has been constructed are shown in a separate text file:
Please cite the following papers if you use the toolbox in your work:
You may contact Dr Melanie Ooi or Dr Ye Chow Kuang through email for any feedback or bugs. Please start your email subject with "ADTree" when you contact the authors.
|Dr Hong Kuan Sok
Monash University Alumni
|Dr Kuang Ye Chow
Associate Head of School (Research Training)
|Dr Melanie Po-Leen Ooi
|Prof Serge Demidenko