A Study of Applying Decision Tree and Artificial Neural Networks on Account Receivable Model of Bad Debt

• Main Authors: I-TSUNG KUO, 郭一聰
• Other Authors: Wei-Ping Lee
• Format: Others
• Language: zh-TW
• Published: 2005
• Online Access: http://ndltd.ncl.edu.tw/handle/14704388300479038715

碩士 === 中原大學 === 資訊管理研究所 === 93 === In the past, companies retrieve cash through selling and marketing activities. Nowadays, because of the extension of credit system which makes companies hard to receive cash immediately. Enterprises have to accomplish the transaction by means of accounts receivable. Theoretically, the shorter the collecting days, the safer and the least possible the accounts receivable turn into bad debt. To prevent companies from the disadvantages of bad debt, the occurrence and precaution of bad debt have become a hot issue problem in the present enterprises. The economic fluctuations cause many punctual debtors default their payments which were thus credited to bad debt. But this kind of bad debt is easier to collect, compared to the other types of bad debts. This study applies decision tree and neural networks technology, based on customers’ background information and the data of accounts receivable after transaction, to form the analytical variables. In addition, to build up an early warning overdue model for the enterprises to analyze overdue accounts receivable, review and reset the credit line; moreover, to decrease the possibility of bad debt. The result of the study found that collecting days after transaction, risk rating, customer classification, franchising time, unemployment rate, transaction times in the past six months are predictable variables in the early warning overdue model using decision tree and Neural Networks technology.