Food safety, nutrition, and public health are actual economic and medical problems. Sweetness is an important feature of food technology. Models for the sweetness of special organic compounds used in the food industry are suggested. The models are built using the CORAL software. New statistical coefficients of predictive potential are studied. These are the index of ideality of correlation (IIC) and correlation intensity index (CII). The effectiveness of using the IIC and CII has been tested in simulated sweetness via Monte Carlo optimization of correlation weights for molecular features extracted from Simplified Molecular Input Line Entry System (SMILES) strings. Both factors have been shown to improve the model’s statistical quality on the calibration and validation sets. However, this is accompanied by a decrease in the statistical quality of the training sets.
Toropova et al. (Thu,) studied this question.