Organisation: Vale – UNSW
Start and Estimated Duration: 21 – Jun – 2018, 28 Months
Summary: Predicting Mineral Product (MP) prices have been a significant and difficult task usually addressed by econometric, stochastic-Gaussian and time series methods. None of these methods has proved suitable to characterise the dynamic behaviour and time-related nature of MP markets. Chaos Theory (CT) and Machine Learning (ML) methods can signify the temporal relations of variables, and their evolution has been used separately to understand better and represent MP markets. CT can determine a system’s dynamics in the form of time delay and embedding dimension. However, this information has often been exclusively used to define the system’s behaviour and not for predicting. Compared to usual methods, ML has better performance for predicting MP prices, due to its capacity for finding patterns governing the system’s dynamics. However, the rational nature of economic complications increases concerns regarding the use of hidden patterns for predicting. Therefore, it is indeterminate if variables selected, and hidden patterns found by ML can represent the economic rationality. In the face of their refined features for representing system dynamics, the separate use of either CT or ML does not deliver the expected realistic accuracy. By itself, neither CT nor ML can identify the primary variables affecting systems, recognise the relation and influence of variables through time, and discover hidden patterns governing systems evolution simultaneously. This project discusses the necessity to adapt and combine CT and ML to obtain a more realistic representation of MP market behaviour to prediction long-term price trends.
Comments (0)