Bayesian Network Design for Crude Palm Oil (CPO) Price Prediction Driven by Fluctuation Patterns and Trends

Abstract

Crude Palm Oil (CPO) becomes the alternative to petroleum because it can be refined into different products, such as biodiesel. The price of crude palm oil can be considered the future direction of the world economy. CPO prices are based on market demand and the oil supply from the CPO producer, where they fluctuate and have a huge influence on economy. The price pattern of CPO requires a variety of factors to predict accurately. It needs a lot of data analytics to predict and respond in quick time to the highly variable CPO market. In this work, a dynamic probability model is proposed to predict CPO prices using the Bayesian rule. The rule is formed to be effective in responding to changeable market demands and supplies. It formulates the approximation factors for indicating the expected price of the CPO for 12 months. The dynamic training data has been modelled to fit the Bayesian rule. The process of Prediction and hypothesizing price fluctuations is identified according to several level of Bayesian rule. It is then used to generate a measure of deviation between the actual and probable prices. Our Bayesian-based prediction pricing model is able to predict the price pattern that are comparable to existing benchmark data and shows a lower standard error of regression. The prediction approach can help the traders have a better analysis of price fluctuations in CPO demand and supply. Using Bayesian probability not only improves prediction rules, but it can also forecast CPO trades in high fluctuation situations.