Evaluating the Efficiency of the Classifier Method When Analysing the Sales Data of Agricultural Products

Abstract

Data classification as a method of input analysis is of the greatest interest and necessity for proper distribution and quality evaluation of agricultural products. The use of classification methods allows predicting whether a selected sample from the data set will fit into a particular class or group, which is necessary for the process of sorting products. This study presents the results of a comparative analysis of high-performance classifiers for assessing the effectiveness of further use in the sorting of agricultural products. The study was carried out utilising the classifiers of k-nearest neighbours, naive Bayesian classifiers, and artificial neural networks for data analysis during apple fruit sorting. It has been established that the greatest accuracy 99% of the results is demonstrated by the classifiers of k-nearest neighbours, but, at the same time, they show the lowest calculation speed (0.47 s). The best performance at any data size (65-100%) is shown by the neural network. A comprehensive review of the features and restrictions of the studied classification algorithms, as well as their applications in various areas of agriculture, has been performed.