N.S. Dwi Agustin, Septiadi Padmadisastra
This paper presents study about the parameter estimation in hidden markov model. The approach is taken from a Bayesian method, there will be two sources of information,there are information from the likelihood function and the prior function. This approach will be applied to daily rainfall data in Darajat, Garut. The numbers of hidden states are used in this paper based on Schmidth and Fergusson's climate classification which are suitable to the local conditions. This classification was obtained three types of division in the period of one year where the condition called wet months when monthly rainfall > 100 mm per month, moist months when monthly rainfall between 100 - 60 mm and the dry months when monthly rainfall <60 mm per month. The process estimation of hidden markov parameters is using Gibbs Sampler algorithm. © 2006-2014 Asian Research Publishing Network (ARPN).
Departement of Industrial Engineering, Widya Mandala Catholic University, Surabaya, Indonesia; Departement of Statistics, Padjadjaran University, Bandung, Indonesia