Forecasting is a discipline that has developed considerably in the last three decades in the operational research and industrial engineering field. According to (Fildes et al. in J Oper Res Soc, 2008), four main forecasting approaches can be distinguished: (1) extrapolation; (2) causal and multivariate methods; (3) judgmental forecasting; (4) computer-intensive methods based on data mining procedures applied to large datasets. Of these approaches, decision tree (DT) (Breiman in Classification and regression trees, The Wadsworth & Brooks/Cole, 1984) and general linear model (GLM) are the most robust and easy-to-use algorithms in the data mining field. Here we test data mining models to improve durum wheat price forecasting. To do so, four stages were followed: problem description based on a real word case study, data collection, data mining prediction and statistical analysis.