Prediction of the crop is a dominant element of agriculture and even for farmers[1]. Currently, India is in the second position in the world for agricultural produce. The main economic sector of India is agriculture, which plays an important role in the growth of the economy[1]. Prediction of a crop is a challenging issue, so comprehensive varieties of crop prediction methods are used. In this paper the essential data is collected from districts of Karnataka and Tamilnadu with many parameters like year, district, area, temperature, rainfall, crop, and yield in tons for the year 2005 to 2016. For crop prediction, methods like K-Means and the J48 algorithm are applied in the existing system for clustering and classification. In the proposed system for clustering, the PAM (PartitionaroundMedoid) is applied and for classification, J48 is applied. A dataset is collected and clustered as stated by the attribute of the PAM algorithm by using the Euclidean formula, calculatingthe distance between the points. In this work tools like eclipse mars.2 and the programming language, Java is applied. The result acquired through an existing algorithm such as K-Means & J48 in comparison with a proposed algorithm like PAM and J48 gives better results in terms of accuracy and time complexity.