English  |  正體中文  |  简体中文  |  Items with full text/Total items : 51510/86705 (59%)
Visitors : 8266047      Online Users : 103
RC Version 7.0 © Powered By DSPACE, MIT. Enhanced by NTU Library & TKU Library IR team.
Scope Tips:
  • please add "double quotation mark" for query phrases to get precise results
  • please goto advance search for comprehansive author search
  • Adv. Search
    HomeLoginUploadHelpAboutAdminister Goto mobile version
    Please use this identifier to cite or link to this item: http://tkuir.lib.tku.edu.tw:8080/dspace/handle/987654321/94205


    Title: 平行化處理在決策樹演算法之應用
    Other Titles: Application of parallel computation on the decision tree algorithms
    Authors: 李智慎;Li, Chih-Shen
    Contributors: 淡江大學統計學系碩士班
    陳景祥
    Keywords: 資料探勘;分類器;平行化;迴歸樹;data mining;classifiers;parallel;CART
    Date: 2013
    Issue Date: 2014-01-23 14:08:31 (UTC+8)
    Abstract: 在資料量急劇增加之下,電腦資料的處理速度成了一項重要的技術,如果能夠將資料處理分析的時間節省下來則可提早一步作預測或判斷,而平行化處理就是一個可以使計算時間縮短的一個方法。而在本篇研究裡就是將資料探勘上常用到的決策樹與平行化作結合,本研究用的是以CART演算法配合兩種平行處理方式,第一個方法為資料平行化,是將資料分作多個等分且以平行化的方式建構多顆決策樹;第二個方法為預測變數平行化,其方法是在決策樹計算最佳變數和最佳切割點時將資料以變數個數作區分,分作多個部份來作平行運算,而在最後模擬結果可以得知速度方面CART使用預測變數平行化方法節省的時間比資料平行化還要來得多,而資料平行化方法則必須要在資料量大於大概6萬以後才會在計算速方面比原始無平行化的CART方法快,且在CPU從2核心上升到4時,時間上面的變化都有很明顯的縮減,兩方法在大型資料下都能縮減決策樹在計算上的時間。
    As the amount of data becomes increasing more and more today, the computation speed of the computer becomes a very important factor. If we can decrease the computing time, we could make the prediction or decision much early than expected. Parallel computation is one of the methods which can decrease computing time. In this paper, we will combine the parallel computation with decision tree technique, and use the CART algorithm as our decision tree model. There are two types of parallel algorithms in this research, 1. Data parallelization: split the data into several parts to build the decision tree in parallel, 2. Variable parallelization: split the predictive variables to do parallelization. In our simulation, we will show that method 2 together with the CART tree algorithm has better performance than method 1. On the other hand, CART with method 1 will be faster than non-parallel CART tree when the size of data amount is over 60,000. Also, when the number of CPU kernels are added from two to four, there is a significant reduction in computing time in both algorithms.
    Appears in Collections:[統計學系暨研究所] 學位論文

    Files in This Item:

    File SizeFormat
    index.html0KbHTML177View/Open

    All items in 機構典藏 are protected by copyright, with all rights reserved.


    DSpace Software Copyright © 2002-2004  MIT &  Hewlett-Packard  /   Enhanced by   NTU Library & TKU Library IR teams. Copyright ©   - Feedback