Hidden Markov Model as the Predictive Time Series Movement of Natural Gas Price

Natural gas is a vital component of energy supply in the world. The price of natural gas as energy commodities has a huge impact on economic growth. Analysis of natural gas price movements and their volatility is very important for world economic actors. Therefore, the price movement of natural gas cannot be observed directly (hidden), so this study focuses on predicting the probability for an increase at natural gas prices within a certain time series data using the Hidden Markov Model (HMM). There are three fundamental steps in order to solve the HMM model, the first is calculating the probability of observation using the Forward-Backward algorithm, the second is determining the hidden state sequence using the Viterbi algorithm, and the third is predicting HMM parameters using the Baum-Welch algorithm. This study uses monthly natural gas prices data for the period of February 2011 until 2021. The results showed that the most optimal observation sequence prediction for natural gas prices is around March 2020, then natural gas prices are predicted to fluctuate around April and May 2021. In general, based on HMM analysis in the long term, there will be an upward trend in the prices of natural gas in the future. This is presumably due to limited resources and increasing demand. These results can be considered in the analysis of sustainability energy for natural gas demand in the future in the aspect of time series price movement.