Comparing Neural Networks, Support Vector Machines, and Naïve Bayes Algorhythms for Classifying Banana Types

Abstract

One of the most significant fruits for human consumption is the banana. Fruit consumption not only promotes health but also lowers the risk of heart disease, stroke, digestive issues, hypertension, some cancers, cataracts in the eyes, skin ailments, cholesterol reduction, and, perhaps most importantly, boosts immunity.The study included secondary data, which is information gathered from online resources like Kaggle. Ten categories of bananas will be identified from the 531 total varieties of bananas used as a train dataset: Ambon bananas, Stone bananas, Cavendish bananas, Kepok bananas, Mas bananas, Red bananas, plantains, Milk bananas, Horn bananas, and Varigata bananas. The development of information technology for image object recognition has become a very intriguing topic along with the rapid advancement of society, and it is undoubtedly directly tied to information data. In order to examine Naive Bayes, Support Vector Machine, and Neural Network techniques for classifying banana types, researchers will use the SqueezeNet Deep Learning model to extract features from photos. The study's findings will provide empirical evidence for the distinctions between each algorithm's accuracy, recall, and precision. Based on the collected results, the Neural Network (NN) method is the best in terms of classification, with accuracy of 72.3%, precision of 72.1%, and recall of 72.3%.