IMPLEMENTASI POISSON HIDDEN MARKOV MODELS DAN EXPECTATION MAXIMIZATION ALGORITHM DALAM PENENTUAN MODEL TERBAIK

Abstract

Geographical conditions in Indonesia which is a tropical area strongly supports the occurrence of a hurricane because of the latitude and longitude location . Hurricane is a meteorological phenomenon and its impact can cause considerable damage and losses. The number of occurrences of hurricanes within a certain time can be predicted and prediction results can be used as the basis for handling the case of a hurricane disaster. If the number of occurrences of hurricanes is an observation process and overdispersed, also the cause of the event is not observed directly and form a Markov chain, then the process of observation and its causes can be modeled with Poisson Hidden Markov. The model is characterized by its predictable parameters using the EM Algorithm. From the model obtained then selected the best model based on the AIC value. Research with PHMM obtained 2 model state as the best model with the smallest AIC value, that is 132,2833