This paper addresses the problem of supplier selection for a hazelnut producer in the agriculture industry under the uncertainty of product quality, with the objective of cost minimization. To mathematically represent the problem, a two-stage stochastic programming model is developed to make decisions both before and after the realization of uncertainty. The first stage involves supplier selection decisions, while the second stage determines the amount of product received from each supplier. After the hazelnuts are produced, if they do not meet quality standards, customers may return the products, causing significant customer satisfaction issues and increased costs. Product quality is assessed based on four features: (I) Aflatoxin, (II) Acidity, (III) Moisture, and (IV) Sifter. To address this problem within the context of intelligent agriculture, four different scenario reduction techniques are employed: (I) Sample Average Approximation (SAA), (II) SAA combined with Machine Learning-1 (SAA + ML1), (III) SAA + ML2, and (IV) SAA + ML3. Each method utilizes the K-means clustering algorithm with specific problem-related characteristics. Computational results indicate that when the SAA approach is combined with K-means clustering and scenarios are reduced based on clustering results, better outcomes are achieved in reducing costs associated with low-quality products. Specifically, higher numbers of suppliers and higher probabilities of generating scenarios from relevant clusters contribute to improved cost reductions.