Pemodelan nilai tukar Rupiah terhadap Dolar Amerika menggunakan Hidden Markov satu waktu sebelumnya

Abstract

An exchange rate of Rupiah to American Dollar has become one of important reference for Indonesian economic movement. The movement of the exchange rate of Rupiah is an event that can occur anytime in a long period and possible to reoccur in the future. If the cause of event is not observed directly and forms a Markov chain, so the pair of the cause and an exchange rate of Rupiah can be modeled by hidden Markov. In this thesis the previous time hidden Markov model is used. This model assumes that the present exchange rate of Rupiah depends on the previous exchange rate of Rupiah and the present and previous cause. Model parameter is estimated by using maximum likelihood method and the calculation uses iterative algorithm expectation maximization (EM). Numerical computation is done by using Mathematica 10. After the parameter model is obtained, then the exchange rate of Rupiah to American Dollar can be estimated. Model accuracy is measured by using mean absolute percentage error (MAPE). Resulted MAPE is 4.48% with one iteration.