Development and validation of empirical models for the prediction of selected antioxidants in stored oranges

Abstract

Abstract: Empirical models were developed to predict some selected antioxidants present in stored oranges under a modified atmosphere.  Three sets of four different types of passive evaporative cooling structures made of two different materials; clay and aluminium were designed and constructed as part of the study.  One set consisted of four separate cooling chambers.  Two cooling chambers were made with aluminium container (cylindrical and rectangular shapes) and the other two were made of clay container (cylindrical and rectangular).  These four containers were separately inserted inside a bigger clay pot inter-spaced with clay soil of 5 cm (to form tin-in-pot, pot-in-pot, tin-in-wall and wall-in wall) with the outside structure wrapped with jute sack.  The other two sets followed the same pattern with interspacing of 7 and 10 cm respectively.  The set with 7 cm interspace served as the control in which the interspace soil and the jute sacks were constantly wetted at intervals of two to four hours depending on the rate of evaporation with water at room temperature.  The other two sets (5 and 10 cm interspaced soil) were constantly wetted with salt solution (Table salt (Nacl)) at the same interval to keep the soil in moist condition.  Freshly harvested matured oranges were used for the experiments and the temperature and relative humidity were monitored daily.  The vitamin A, C and E, bacterial and fungal counts of this produce were determined at intervals of three days for a period of 21 days.  Mathematical models (using essential regression software package) were developed to predict the vitamin A, vitamin C, and vitamin E contents of the stored oranges at various conditions considered in the study.  The existence and sufficiency of the regression models given in the equations were also examined using the analysis of variance (ANOVA) of the multiple regression models.  The models were found to be at 5% level of significant.  The models were validated using pair-wise T-test and the results of the pair-wise shows that there is no significance difference between the mean of observed and the predicted for all the models developed.  The R_adj² value obtained were 82.43%, 86.63% and 76.48% for vitamin A, vitamin C, vitamin E contents respectively for the stored oranges.

Keywords: model, validation, predicted, structures, oranges