THE DEVELOPMENT OF A MODEL TO STUDY RANDOMIZED CLASSIFIER ENSEMBLES
|
Ensembles are a combination of multiple base models. The final classification depends on the combined outputs of individual models. Classifier ensembles have shown to produce better results than single models, provided the classifiers are accurate and diverse. Ensembles are very popular because of their excellent classification accuracy. Researchers have shown that ensembles can also be used in feature selection, missing data problems. These are being used in almost every field, bioinformatics, text mining, finance etc. Decision trees are popular classifiers because of its simplicity. Several methods have been proposed to build decision tree ensembles. Randomization is introduced to build diverse decision trees. Random Forests, Bagging, Extremely Randomized Trees and Random subspaces are some of the popular ensemble methods. Excellent performance of ensembles based on randomization leads to many theoretical studies to understand their performance. However, there is no unified model that can exactly predict the behavior of decision tree ensembles. In this project, we will develop a model that can be used to study all the decision tree ensembles based on randomization. That will be helpful in developing new ensemble methods.
|