Identification of Dry Bean Seeds Using PSO Feature Selection Technique

Automatic classification of seed varieties is critical for seed producers to maintain the purity of a variety and crop yield. Seed varieties of premium quality are more costly since their ability to increase productivity and profit margins. Dry bean seeds are one of the most widely used seed varieties in Turkey. Therefore, the classification of dried beans is vital both for the purity of the product and for production and marketing. Learning-based classification and optimization approaches are increasingly being employed in all fields of research as a result of advancements in computer and machine learning techniques. The aim of this study is to classify the types of dry beans, that are, Barbunya, Dermason, Cali, Bombay, Sira, Seker and Horoz. The dataset used in the study contains data consisting of sixteen attributes belonging to seven dry bean varieties. Firstly, particle swarm optimisation (PSO) method from metaheuristic algorithms for the characteristics of the data set feature selection process was carried out by using the feature selection algorithm. Secondly, the selected features were identified using machine learning techniques such as decision tree (DT), Naive Bayes (NB), support vector machine (SVM). The highest identification accuracy for the selected features was SVM with 93.03%.