Bayesian optimized ANN model on Vegetable Oil based PHA Production
Description
The data set is biomass growth and PHA production from multiple batch studies (endpoint analysis) differing on inoculum, sucrose-oil ratio, and time of phase shift. The data is used to develop ANN model with hyperparameter optimization. The finalized model is analyzed for SHAP, LIME for explanation.
Files
Steps to reproduce
mydata_raw.mat is the data set to start. a0_Fitlm_ML_model_select.m is for ANOVA analysis. For ANN modelling, the sequence of files is to be executed as a1_fitrnet_bayesian_Opt_PHA_ori.m to have Bayesian optimization. The selection of best model and graphical analysis is in a2_fitrnet_bayesian_Opt_Analysis.m The lime and SHAP analysis is done using a5__LIME_Shaply.m, and visualization using a6_LIME_Shaply_finalplot.m. The different mat files are created from executing .m file scripts in sequences.