Improving time series forecasting using information fusion in local agricultural markets

This research explores the capacity of Information Fusion and Data Mining to extract knowledge about associations among agricultural products and achieve better predictions for future consumption in local markets in the Andean region of Ecuador. This commercial activity is performed using Alternative Marketing Circuits (CIALCO), seeking to establish a direct relationship between producer and consumer prices, and promote buying and selling among family groups. The time-series forecasting, presented as a machine learning formulation, is enhanced with multi-variate predictions based on association rules extracted from transactions data analysis. These transactional data are used to learn the best association rules between products sold in different local markets, knowledge that allows the system to gain a significant improvement in forecasting accuracy by including these variables in multi-variate forecasting models. In the results we see that, from establishing best association rules valid in the original dataset, we can achieve a considerable improvement in prediction accuracy, validated with independent test subsequences of agricultural products using non-linear regression techniques including neural networks with a varying number of hidden layers.