Multiclass Date Fruit Prediction Using SVM and Logistic Regression with OVO and OVR

Abstract

Multiclass classification plays a crucial role in various AI applications requiring simultaneous recognition of multiple classes. This research investigates the application of multiclass classification techniques to predict the species of date fruits using Support Vector Machine (SVM) and Logistic Regression (LR) algorithms with One-Versus-One (OVO) and One-Versus-Rest (OVR) strategies. The dataset comprises 898 samples with 35 elements and 7 distinct labels. The experimental results demonstrate the superiority of SVM over LR, achieving the highest accuracy of 93.68% with the OVR strategy. Moreover, OVR outperformed OVO in both algorithms, showcasing its efficacy for multiclass problems. These findings offer valuable insights for date fruit prediction and further advance the state of multiclass classification techniques in AI applications.