Inter related metal price prediction based on eABC-LSSVM

This study investigates the predictability of two inter related metal prices, namely gold and palladium using an Enhance Artificial Bee Colony (ABC) and Least Squares Support Vector Machine (LSSVM). The idea is to use eABC as an optimizer to automatically adjust the value for LSSVM's hyper parameters, namely regularization parameter, γ and kernel parameter, σ². As to prevent the premature convergence, a mutation strategy is incorporated to improve the standard ABC. This study evaluates the effectiveness of the proposed model by comparing its performance against the one produced by the standard ABC. By using time series data of 2008-2011, the prediction performance was evaluated based on Mean Absolute Percentage Error (MAPE). Analysis of the experimental results indicate that LSSVM model achieves better prediction performance when hybrid with eABC.