Performance Comparison of Machine Learning Algorithms for Identification of Physiological Maturity of Pineapple using Optical Property

Abstract

Pineapple is important fruit of Thailand which is consumed in its fresh state or in processed products. Typically, harvested dates affected to quality of pineapple fresh. Identification of pineapple harvested dates at raw material receiving state in factory is very difficult. This research aims to determination of appropriate machine learning algorithm for Identifying maturity of pineapple using optical property. Color of pineapple fruits and fresh was measured by portable colorimeter on CIE system (L*, a* and b* values). The ten algorithms were fit to the training set including naive Bayes (NB), linear discriminant analysis (LDA), K-nearest neighbor (KNN), support vector machine (SVM), artificial neural network (ANN), logistic regression (LR), decision tree (DT), random forest (RF), gradient boosting (GB) and adaptive boosting (AB). The best model for pineapple fruit were established from ANN while DT showed highest performance for pineapple fresh. The accuracy of ANN and DT for fruit and fresh models were 83 and 92% respectively. This finding point is novel technique for identification of pineapple according to harvested dates which it can apply to quality control and assurance in pineapple industries.