Articles
A comprehensive quality evaluation model for watermelon based on textural characteristics, physical, and nutrition indexes
Article number
1411_26
Pages
261 – 274
Language
English
Abstract
The soluble solid content of watermelon is a main factor affecting its commodity.
Except for the soluble solid content, the appearance, mouth feel, and nutrition ingredients were all closely related to the comprehensive quality of watermelon.
Therefore, the sensory scores, textural characteristics, and nutrition of watermelon were evaluated to construct a comprehensive quality model.
The multiple linear regression, principal component regression, stepwise regression, ridge regression, and partial least squares regression algorithm were employed to construct a comprehensive quality model with the ‘L600’ cultivar (Citrullus lanatus L.) as material.
Specifically, the correlation analysis showed that 11 indexes, including pulp hardness, pulp brittleness, moisture content, L* value, a* value, titratable acid, soluble solids, sugar-acid ratio, pH value, vitamin C, and lycopene, were closely correlated with the sensory scores.
The 11 indexes were regressed by the mentioned algorithms.
The partial least squares regression model showed the highest matching degree (comprehensive quality evaluation y = 403-1.80 × pulp hardness – 8.18 × pulp crispness – 3.69 × moisture content – 0.11×L* value + 0.06 × a* value – 0.11 × titratable acidity + 3.90 × soluble solids + 2.67 × sugar-acid ratio – 12.14 × pH value + 1.00 × vitamin C + 0.03 × lycopene) with Rc2 of 0.945, RMSEC of 6.241, Rp2 of 0.943, RMSEP of 3.526, RPD of 4.171. The RPD of the partial least squares regression model was improved by 13.4% compared with that of the principal component regression model.
Except for the soluble solid content, the appearance, mouth feel, and nutrition ingredients were all closely related to the comprehensive quality of watermelon.
Therefore, the sensory scores, textural characteristics, and nutrition of watermelon were evaluated to construct a comprehensive quality model.
The multiple linear regression, principal component regression, stepwise regression, ridge regression, and partial least squares regression algorithm were employed to construct a comprehensive quality model with the ‘L600’ cultivar (Citrullus lanatus L.) as material.
Specifically, the correlation analysis showed that 11 indexes, including pulp hardness, pulp brittleness, moisture content, L* value, a* value, titratable acid, soluble solids, sugar-acid ratio, pH value, vitamin C, and lycopene, were closely correlated with the sensory scores.
The 11 indexes were regressed by the mentioned algorithms.
The partial least squares regression model showed the highest matching degree (comprehensive quality evaluation y = 403-1.80 × pulp hardness – 8.18 × pulp crispness – 3.69 × moisture content – 0.11×L* value + 0.06 × a* value – 0.11 × titratable acidity + 3.90 × soluble solids + 2.67 × sugar-acid ratio – 12.14 × pH value + 1.00 × vitamin C + 0.03 × lycopene) with Rc2 of 0.945, RMSEC of 6.241, Rp2 of 0.943, RMSEP of 3.526, RPD of 4.171. The RPD of the partial least squares regression model was improved by 13.4% compared with that of the principal component regression model.
Publication
Authors
Shanshan Li, Yingguo Lyu, Hongxu Wang, Mengqing Lu, Fan Sun, Xueshan Wen, Chao Zhang
Keywords
watermelon, comprehensive quality, sensory evaluation, nutrition, textural properties, regression model, model verification
Groups involved
Online Articles (37)