Using case-based reasoning to evaluate real estate pricing

• Main Authors: Chia-Chun Chen, 陳佳駿
• Other Authors: Chin-Ho Lin
• Format: Others
• Language: zh-TW
• Published: 2006
• Online Access: http://ndltd.ncl.edu.tw/handle/09817143574089586898

碩士 === 國立成功大學 === 資訊管理研究所 === 94 === Accurate measuring of real estate price is a very important issue in housing market. However, there are many dynamic factors that will influence the quality of house-price evaluating. On the other hand, the traditional evaluating tools are often very difficult to learn and use. Therefore, sales comparison approach, using the price of neighboring house to estimate the target value, is often adopted in the evaluating process because of its simplicity and efficiency. CBR (Case-Based Reasoning) is a well-known methodology of inference system in the decision area. CBR stores the historical cases in the case base, and retrieve the similar case to provide the past solution when dealing a new case. The system works quite like the way that human solving new problem and its principle also correspondences to the sales comparison approach. This paper represents an approach that combines the sales comparison approach and CBR system to provide evaluating value of a housing case. According to the past studies, the attribute weight of a case, usually defined by experts’ experience, may affect the result of case selection in a CBR system. In the prototype system of this paper, the heuristic method, genetic algorithm (GA), is adopted to adjust these attribute weights reasonably and consistently. Finally, the system use a similarity function to convert these attributes values to a suitable evaluating price for decision support. In the following experiment with the Taipei housing cases, the system can provide good estimation of these cases.