English  |  正體中文  |  简体中文  |  全文筆數/總筆數 : 49521/84606 (59%)
造訪人次 : 7582829      線上人數 : 75
RC Version 7.0 © Powered By DSPACE, MIT. Enhanced by NTU Library & TKU Library IR team.
搜尋範圍 查詢小技巧:
  • 您可在西文檢索詞彙前後加上"雙引號",以獲取較精準的檢索結果
  • 若欲以作者姓名搜尋,建議至進階搜尋限定作者欄位,可獲得較完整資料
  • 進階搜尋
    請使用永久網址來引用或連結此文件: http://tkuir.lib.tku.edu.tw:8080/dspace/handle/987654321/87504


    題名: A credit cardholder behavioral scoring model using residual correction Fourier GM (1,1)
    其他題名: 灰色傅立葉行為評等模式之建構
    作者: 林哲敏;Lin, Che-Min
    貢獻者: 淡江大學管理科學學系碩士班
    陳怡妃;Chen, I-Fei
    關鍵詞: 行為評等;灰色系統理論;灰色傅立葉;馬可夫鏈;貝氏機率;Behavioral Scoring;grey system theory;Grey Fourier;Markov Chain;Bayesian
    日期: 2012
    上傳時間: 2013-04-13 11:17:11 (UTC+8)
    摘要: 本研究之目的是有鑒於金融機構中所隱藏之潛在信用風險,故藉由縮短顧客還款行為之觀察期,以建構預測顧客未來盈利能力之行為評等模式。藉由灰色理論善於處理短期資料之特性,我們建立GM(1,1)模型來縮短預測觀察期並驗證其用在分類資料的適用性,企圖降低銀行可能潛在的信用風險。然而,事實上GM(1,1)乃常用於預測面問題之解決,鮮少用於處理分類之相關議題上,導致其預測準確度不如預期,因此本研究將GM(1,1)模型結合具有週期性函數特性的傅立葉轉換,將其應用於殘差修正上,以提高模型的預測精確度。再者,由於傳統上馬可夫鏈被廣泛地應用在信用評等及行為評等上,故本研究利用馬可夫鏈模型之預測準確率視為灰色傅立葉模型之參考基準。最後,本文針對GM(1,1)模型,FGM模型,馬可夫鏈模型以及BGM(張雯琪,2011)等四個模型進行相關之比較,結果顯示FGM模型以及BGM模型皆有優異的預測準確率,且成功地將顧客還款行為之觀察期縮短為少於十期,此舉對於新申請帳戶之顧客,亦可達到快速制定授信決策之效,如此便能針對既有客戶提供更適切的服務,進而增加銀行之利潤。
    The purpose of this study is to construct the behavioral scoring model of predicting customer future profitability individual by shortening the period of observation his/ her payment profitability. Firstly, We construct GM(1,1) model to test the applicability of short-observation credit prediction associated with classification problems. Then, we proposed Fourier residual grey modification model (FGM) to improve the predictive accuracy. Next, we use Markov chain, a widely applied traditional method for solving credit/behavioral scoring problem, to provide a reference level of prediction accuracy. Finally, after comparing to GM, FGM, MC and GBM developed by Chang (2011), we find that the FGM(1,1) model and Bayesian grey model have outstanding performance of prediction accuracy. We find that the FGM(1,1) model and Bayesian grey model have outstanding performance of prediction accuracy and shorten the observation periods for less than 10 observations, successfully. This study delivers a managerial insight that the proposed model enables banks to take effect of the quick credit decisions, and then the financial institute can design appropriate marketing portfolios management based on the more accurately predicted status of customers future profitability.
    顯示於類別:[管理科學學系暨研究所] 學位論文

    文件中的檔案:

    檔案 大小格式瀏覽次數
    index.html0KbHTML176檢視/開啟

    在機構典藏中所有的資料項目都受到原著作權保護.

    TAIR相關文章

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