| Summary: | 碩士 === 國立臺灣科技大學 === 資訊管理系 === 92 === After 1998 Taiwan Futures Exchange was established, and provided the futures transactions for compatriots. Now there are average 20,000~30,000 transactions on each trading day. This indicated that the futures market is in great demand.
Besides, computer science and artificial intelligence can assist us to process mass information, especially decision support is very prevalent. This paper used rough sets theorem and fuzzy theorem to forecast the rank of TX price fluctuation.
There are four methods of rough sets and fuzzy theorem adopted in this study:
1. classical rough sets theorem
2. rough fuzzy sets theorem
3. rough sets theorem with dynamic adjust rank
4. improved rough sets theorem
The empirical results consist of the following:
1. The four methods average forecast accurate rate was not bigger than 70%, the possible reason is that TX price fluctuation was affected by many factor, so we can’t correctly forecast it.
2. The fourth method can forecast TX price fluctuation direction average 70% accurate rate. The first and third methods can forecast TX price fluctuation average 20% better than probability.
3. The paper suggested a new data structure to solve rough sets theorem implementation that can increase the forecasting performance and be used easily.
|
|---|
