Prediction of Yields Fluctuations on Indonesian Government Bonds Using Classification Model Based on Rough Sets and Fuzzy Rough Sets

Investment is one of the important factors for the country's economic growth. One of the investment instruments is the yield on Indonesian government bonds which have changed occasionally. Predictions of its fluctuations are useful for determining bond investment strategies in the future. In this paper, we utilized quick reduct algorithm based on rough sets and fuzzy rough sets to select the best features incorporated with CN2 algorithm to predict the fluctuations of the yield values. The prediction was made in the classification of the closing price of Indonesian government bond yields which would increase, decrease, or tend to be constant the next day. Based on simulation results, even though the number of rules produced was smaller, the CN2 algorithm based on fuzzy-rough sets has better accuracy than rough sets in predicting fluctuations of Indonesian government bond yields.