The integration of information and digital communication tools in the agricultural sector has led to the emergence of precision agriculture. This field now utilizes Internet of Things (IoT) technologies, geographical data, and both historical and real-time information, all of which have the potential to transform farming into a smart, technologically advanced enterprise. IoT and machine learning can synergize to create fully intelligent systems. In precision agriculture, clustering is often employed to extract knowledge for optimizing crop productivity, irrigation control, and food quality and safety monitoring. This paper proposes an unsupervised clustering algorithm based on the concept of Markov Blanket Approximation to group agricultural data collected in an IoT environment. The algorithm is evaluated using a dataset extracted from the Meteoblue platform, and the results demonstrate its effectiveness compared to the K-means algorithm.