Multi-objective Optimization using ANN and Quadratic Model
Description
This dataset contains CCD result data for pullulan extraction using the solvent method and codes for developing linear regression and artificial neural network (ANN) models to optimize the extraction process (i.e., solvent-to-broth ratio, pH, and incubation time). It includes scripts for performing sensitivity analysis, desirability testing, and multi-objective optimization for three responses (i.e. pullulan recovery, sucrose equivalent, and protein impurities). The package also provides code to generate response surface methodology (RSM), an ANN residual plot to evaluate model accuracy, and Pareto front analysis to identify optimal trade-offs between competing objectives.
Files
Steps to reproduce
Two different modeling approaches for CCD experimental data are presented. One folder is named Fitlm - on linear regression model, and the other one ANN-based model. In both folder, the scripts are alphabetically arranged to be sequentially used in the exercise, The first script is for model development, the next script intends to optimize the responses individually, subsequently, multiobjective optimization in GA, and pareto optimal solution is rendered.