English  |  正體中文  |  简体中文  |  Items with full text/Total items : 63184/95884 (66%)
Visitors : 4552880      Online Users : 292
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: https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/74243

    Title: 貝氏灰行為評等模式之建構
    Other Titles: Bayesian grey classification in behavioral scoring
    Authors: 張雯琪;Chang, Wen-Chi
    Contributors: 淡江大學管理科學研究所碩士班
    Keywords: 行為評等;灰色系統理論;貝氏機率;Behavioral Scoring;grey system theory;Bayesian
    Date: 2011
    Issue Date: 2011-12-28 18:18:00 (UTC+8)
    Abstract: 金融機構於信用循環(credit cycle)之管理程序中,潛在壞帳之預警及防範,是其重要課題之一,而縮短對顧客信用/還款行為評等之觀察期,並有效預測顧客未來還款行為,更可俾金融機構掌握預先示警之時效,然為縮短對顧客之觀察期並有效預測其未來還款行為狀態,本研究結合灰色系統理論與貝氏機率建構三階段之「貝氏灰行為評等模式」,援引灰色預測處理短期資訊之長處,企圖降低蒐集顧客過往信用交易記錄所耗費之時間與倉儲等成本,然又於文獻中,GM(1,1)乃常用於預測面問題之解決,鮮少用於處理分類相關議題,故藉由grey cluster分析將由GM(1,1)所獲致之連續數值轉換成具分類效用之各類別機率,後更輔以貝氏事後機率修正,藉此提高模式之鑑別率。
    為驗證模式之適用性,採用金融個案公司所提供之信用卡顧客資料進行實證,結果顯示,本研究提出之貝氏灰行為評等模式,於兩期測試樣本之鑑別率分別為92.98%與88.76%,確有優於單獨使用GM(1,1)之27.68%與26.75%及其結合grey cluster之二階段灰模式30.97%與30.78%之鑑別表現。
    For financial institutes, an advance warning and prevention of potential delinquency are very crucial to their credit risk management. Furthermore, shortening the observation periods of customers’ payment behaviors without compromising accuracy of predicting customers’ future credit status is referred to a contributing factor to the enterprise time-based core competence.
    Taking advantages of grey theory superiorly able to tackle subtle small sample problems and Bayesian learning process to correct prediction errors, this study aims to propose a 3-stage credit payment behavioral scoring model based on the underlying rationale that the resulting continuous numbers from the 1st stage grey credit prediction are transferred into the corresponding probabilities to predicted classes by the 2nd stage grey clustering and the predicted classes are consequently revised by the posterior probabilities of Bayesian theorem at the 3rd stage.
    The proposed model is constructed on 10 observation periods as training samples and 2 performance periods as testing samples. Also its applicability is examined on a real-life dataset of 2677 credit cardholder accounts, each of which contains 12 historical payment periods. Every cardholder in each period was ranked as 1, 2, and 3 in accordance with ascending profitability of his/her payment behaviors. This empirical results show that the accuracies of this proposed model on 2 testing samples respectively are 92.98% and 88.76%, strikingly outperforming grey prediction GM(1,1) model of 27.68%, 26.75%, and 2-stage grey classification model of 30.97%, 30.78%.
    Appears in Collections:[Department of Management Sciences] Thesis

    Files in This Item:

    File SizeFormat

    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