Machine learning model for adsorptive separation of pigment from polymer
Description
All datasets and source codes used for PLS model development and quantification of polymer and conjugated pigment, biochar-based pigment adsorption experimental data with machine-learning modeling and optimization, and raw data with scripts used for figure generation and plotting.
Files
Steps to reproduce
The folder "a0_Developed_PLS_model" contains the subfolder "PLS_codex," which has the basic PLS functions and calibration datasets (X, Y, and wavelength). The remaining scripts in this folder are for full-spectrum PLS modeling, quantification of experimental and validation data, and prediction interval calculation for PLS by the three-term expression method. The folder "a1_Synthetic_fitrnt" contains the adsorption experimental data and codes for Bayesian ANN model development using synthetic data, model selection, GA min and max calculation, desirability-based multi-objective optimization, LIME and SHAP explainability analysis, and jackknife-based prediction interval calculation with figure generation codes. the folder 'a3_PLS_FTIR_figure' to reproduce other figures, including spectral plots, factor loading, VIP score, regression, and residual plots of PLS; FTIR and UV–Vis spectra, point of zero charge plots, and biochar reusability plots.
Institutions
- Karunya UniversityTamil Nadu, Coimbatore