Deterministic forecasting model of fuzzy time series with unsupervised interval partitioning

• Main Authors: Su-Yu Lin, 林書玉
• Other Authors: Sheng-Tun Li
• Format: Others
• Language: en_US
• Published: 2007
• Online Access: http://ndltd.ncl.edu.tw/handle/71136796560786494446

碩士 === 國立成功大學 === 資訊管理研究所 === 95 === With the fast growth of information technology, the issue of how to predict through scientific computation and information analysis becomes crucial. In addition, more accurate and efficient forecast can support the decision-making. In this study, we propose a two-factor time-invariant forecasting model, which is more efficient and can controll uncertainty. Moreover, concerning the affect of the interval partitioning, we combine the forecasting model with fuzzy c-means algorithm to fuzzify the historical data. The data of daily average temperature and average cloud density from June to September, 1996 in Taipei are experimented for performance evaluation. A simple Monte Carlo simulation is performed to achieve the true performance of the model approximately. The proposed model achieves better forecasting performance when being compared in both modeling and forecasting accuracy with other extant researches.