Premium
Combination of unsupervised and supervised models to predict the maturity of peaches during shelf‐life
Author(s) -
Zhong Yuming,
Bao Yao,
Ye Jiaming,
Liu Jianliang,
Liu Huifan
Publication year - 2021
Publication title -
journal of food processing and preservation
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.511
H-Index - 48
eISSN - 1745-4549
pISSN - 0145-8892
DOI - 10.1111/jfpp.15624
Subject(s) - maturity (psychological) , mathematics , shelf life , multivariate statistics , capability maturity model , statistics , horticulture , biology , food science , computer science , psychology , developmental psychology , software , programming language
Maturity is one of the important factors affecting peach ( Prunus persica L. Bastsch ) quality and shelf‐time. In this study, a combination of unsupervised and supervised models was employed to predict the maturity period of peach using 178 species from seven areas in China. Three types of nonparametric tests were used to analyze the data. The results emphasized that maturity may not be influenced by a single factor. Furthermore, PLS‐PM analysis emphasized that the color module was the most important factor during maturity. As a result, k‐means and random forest (RF) models were successfully established to predict the maturity of peaches during shelf‐life; the accuracy was 97.15%, and the kappa was 0.94. In double‐blind tests, the RF model achieved 100% classification over the maturity stages of peaches, and 81.25% confirmed the optimum maturity level, verifying our model. This study provides useful information on the maturity‐associated rapid analysis of peaches during their shelf‐life. Novelty impact statement In total, 178 species of peaches, obtained from seven areas in China, were analyzed. Multivariate analyses were conducted to determine peach maturity. PLS‐PM analysis emphasized the color module as affecting maturity and random forest model reaches 100% classification over the peach maturity stage.