Parallel Decision Tree Algorithm Based on Combination

Author

Wenlong, Li ; Changzheng, Xing

Author_Institution

Electron. & Inf. Eng. Sch., Liaoning Tech. Univ., Huludao, China

Volume

1

fYear

2010

fDate

16-18 July 2010

Firstpage

99

Lastpage

101

Abstract

Early ID3, C4.5, CART and the other decision tree algorithms are no longer met the situation of massive data analysis for the time being. Those algorithms has the same limitations that they can not handle the updated data sets dynamically and the decision tree generated by these algorithms need to be purned. These weaknesses limit the use of the above-mentioned algorithms. So a novel parallel decision tree classification algorithm based on combination (PCDT) is put forward in this paper. This algorithm has the excellent features that it can be updated when the data set is renewing and it is scalable and no pruning. It has been proved by the experiment that the PCDT algorithm has higher classification accuracy and is easy to parallel.

Keywords

data analysis; decision trees; parallel algorithms; pattern classification; C4.5; CART; ID3; PCDT algorithm; combination classification; massive data analysis; parallel decision tree classification algorithm; Algorithm design and analysis; Classification algorithms; Data mining; Decision trees; Heuristic algorithms; Program processors; Training; ID3; combination classification; decision tree; knowledge integration; parallel;

fLanguage

English

Publisher

ieee

Conference_Titel

Information Technology and Applications (IFITA), 2010 International Forum on

Conference_Location

Kunming

Print_ISBN

978-1-4244-7621-3

Electronic_ISBN

978-1-4244-7622-0

Type

conf

DOI

10.1109/IFITA.2010.115

Filename

5635178