HIDDEN MARKOV MODELS-BASED FORECASTING MODEL FOR CRUDE OIL PRICE WITH FUZZY TIME SERIES

• Main Author: Syakina Tikita Farne, Pocut
• Format: Theses
• Language: Indonesia
• Subjects: Matematika
• Online Access: https://digilib.itb.ac.id/gdl/view/33933
• Institution: Institut Teknologi Bandung

Nowadays, crude oil is one of the main fuel used in the world and one of the most crucial commodities for industries and people. The fluctuations of crude oil price cause one to make a right investment decision. So, the crude oil price forecasting is a must for oil producer or consumer. There are two aspects that should be considered in this problem. The first one is the number of production of crude oil as the observed aspect, and the second one is crude oil price as the unobserved (hidden) aspect. Assuming that crude oil price was affected by the number of crude oil production at the same time and crude oil price on one day was affected by crude oil price one day previously, Hidden Markov Models (HMM) is used on forecasting process. Crude oil price and production data are expressed as Fuzzy Time Series (FTS) since it can deal with vague and incomplete data. Crude oil price and production are divided into the same length intervals and denoted linguistically by hidden states (price) and observable states (production), respectively. Emission matrix which is one of HMM’s parameters is used as fuzzy relation matrix to give the connection between HMM and FTS. Through the modification of FTS’s defuzzification process some forecasting results are simulated.