Development of binary models for prediction and optimization of nutritional values of enriched kokoro: a case of response surface methodology (RSM) and artificial neural network (ANN)

Abstract Kokoro as a broadly acknowledge maize snack is being consumed every day by both the children and grown-ups, but it is characterized by the low protein content required for survival. The blending of maize flour (MF), sesame flour (SF) and moringa flour (mF) to enhance the nutritional values of kokoro was optimize with response surface technique (RSM) and artificial neural network (ANN). MF, SF and mF were mixed at diverse proportion and the optimal blending ratio was gotten using D-optimal design method. The protein and carbohydrate actual contents were compared with their predicted values using RSM and ANN models. The performance of the developed RSM and ANN models were validated with coefficient of determination (R2) and mean square error (MSE). The optimal blending ratio of MF: SF: mF was 54.11: 37.06: 8.83. The optimal blending ratio gave 25.53% of protein content and 45.99% of carbohydrate content. The statistical analysis of the experimental data obtained using different statistical techniques shows that regression models by RSM gave R2 of 0.999 for protein yield and 0.983 for carbohydrate yield while ANN gave R2 of 0.999 with MSE 9.24184 × 10−1. Therefore, it can be concluded from the results that both the RSM and ANN gave good prediction of the model.