The quality of edible oils can be deteriorated by the
oxidation
of unsaturated fatty acids. Hence, it would be beneficial to accurately
predict oxidative stability using readily available data, such as
fatty acid composition and antioxidant content. However, the oxidation
of edible oils involves complex radical chain reactions and is therefore
difficult to model. The present work employed multivariate analysis
incorporating chemiluminescence data to develop predictive models
for the oxidation induction period of oils. Predicted values were
then compared to experimental values obtained from Rancimat measurements
of 13 different oils. A linear regression analysis provided a higher
adjusted coefficient of determination in the case in which chemiluminescence
data were included. This model was also able to accurately predict
the stability of oil types not used in the model construction.