Improving the accuracy of intercellular space volume of citrus prediction by artificial neural networks

A. Dirpan, Y. Hikida, H. Chiba, G.K. Aji, A.N. Rahman
The intercellular space volume (Vin) is an important factor in studying gas exchange in fresh produce. In addition, quality of fresh citrus fruits is greatly affected by internal concentrations of CO2 and O2, which are influenced by the Vin. A purpose of this study was to improve the accuracy of Vin prediction in citrus fruits. To improve the accuracy, the study employed artificial neural networks method, in which mass (m), citrus volume as spherical assumption (Vs) and real density (Pt) were used as inputs. The results showed that the determination coefficient (R2) in the validating model increased from 0.860 to 0.981. Furthermore, a paired t-test (0.257) showed that the predicted value and measured value were not significantly different at the 5% level (pKLEINERDAN0.05). This finding indicates that the accuracy in predicting Vin can be improved by an artificial neural network.
Dirpan, A., Hikida, Y., Chiba, H., Aji, G.K. and Rahman, A.N. (2017). Improving the accuracy of intercellular space volume of citrus prediction by artificial neural networks. Acta Hortic. 1179, 37-44
DOI: 10.17660/ActaHortic.2017.1179.7
https://doi.org/10.17660/ActaHortic.2017.1179.7
citrus, prediction, intercellular space volume, artificial neural network
English

Acta Horticulturae