資訊科技的日新月異讓資料的產生和收集技術有急劇地發展，資訊技術已經被廣泛應用到組織和政府中，再加上資料庫能力的提升，讓資料產生爆發性的成長，資料量早遠超過人類能直接分析的能力，因此如何能有智慧且自動的將資料轉換成為有用的資訊及知識，便成為資料庫應用的前瞻目標，而資料探勘（Data Mining）就因此逐漸地成為一個重要的研究領域。 本論文主要是以約略集合（Rough Set）的方法，及資料探勘技術中的關聯規則的方法為基礎，發展出一個以RSFP（Rough set and FP Growth tree）為方法論的模式。此模式先對所有的屬性進行篩選，只剩下最重要的屬性，而後再進行資料探勘以得到最後的關聯規則。而在本論中，我們以醫院申報資料為例，將重要的申報屬性，透過約略集合理論而留下來，對於不重要的屬性，則依據此理論去除之。 本研究利用約略集合作為篩選屬性的重要理論，然後近一步用FP Growth tree演算法來探勘出關聯規則，且利用約略集合和FP Growth tree方法的結合，應用在健保醫療費用資料上。而在研究的結果上，加速了屬性的篩選與探勘的效率，且尋找出院方有興趣之規則。 The innovation of information technology has increased the development of information production and gathering. It is widely used in many companies and government agencies. With the increasing ability of database, information has gathered with rapid pace. The volume of information has exceeded the ability of human direct analysis. Therefore, how to smartly and automatically transfer data into useful information and knowledge becomes a very pioneering goal of data application. As a result, data mining gradually becomes important. The project adapts Rough Set and Association Rules of data miming to develop a model based on RSFP ( Rough set and FP Growth tree). The model sifts all attributes out to leave only t he most important ones. Then it comes to data miming which is used to get the last association rules. In the project, we take the example of reported data from the hospital. We sift the minor attributes out by the theory of rough sets to gather the most important attributes. This project is an important theory to sift out attributes by means of rough sets. Furthermore, I use FP Growth tree to mine association rules. We also apply the combination of both rough sets and FP Growth tree to health insurance fees and data. The result heightens the efficiency of sifting and miming of attributes and also helps the hospitals find out what rules they are interested in.
