An ensemble strategy for forecasting the extra-virgin olive oil price in Spain

Time series prediction is one of the key tasks in data mining, especially in areas such as science, engineering and business. It is possible to distinguish between fundamental analysis and technical analysis while dealing with time series in the business area. Fundamental analysis takes into account different exogenous variables such as expenses, assets or liabilities. Technical analysis summarizes information using technical indicators such as momentums, moving averages or oscillators. The most influential exogenous variables and technical indicators for the olive oil price have been already identified in previous studies. The objective of the present paper is to propose an ensemble strategy, based on dividing this set of exogenous variables and technical indicators into subsets of features for the base models. These base models use CO2RBFN, a cooperative competitive algorithm for RBFNs, as learning algorithm. The obtained results show that the ensemble strategy outperforms both the base models and other classical soft computing methods.