TMM-Generated Multilayer SPR Sensor Dataset and XGBoost Predictions for Arsenic Detection

Dataset Title:TMM-Generated Multilayer SPR Sensor Dataset and XGBoost Predictionsfor Arsenic Detection Associated Paper:Anik, N. D., and Imam, K. T. "Machine Learning-Assisted PerformancePrediction and SHAP-Based Analysis of a Multilayer SPR Sensor forArsenic Detection." Manuscript ID: NXMATE-D-26-03118. File Contents:Sheet 1 — TMM_Dataset_1244samples Complete preprocessed dataset of 1244 valid samples. Inputs (4 columns): d_MgF2_nm, d_Cu_nm, d_Ni_nm, d_ZnSe_nm (nm) Outputs (15 columns): Sensitivity, FWHM, FoM, DA, QF at arsenic concentrations of 0.1%, 0.5%, and 1.0%. Generated using Transfer Matrix Method (TMM) at wavelength 632.8 nm for the S-FPL53/MgF2/Cu/Ni/ZnSe multilayer structure. Sheet 2 — XGBoost_Predictions_TestSet Held-out test set (374 samples) with true TMM values and XGBoost predicted values side by side. Columns: 4 input features + 15 True_ columns + 15 Predicted_ columns. Model: Tuned XGBoost (n_estimators=300, max_depth=8, learning_rate=0.1, colsample_bytree=1). Units: Thickness: nm Sensitivity: deg/RIU FWHM: deg DA: deg-1 QF: dimensionless FoM: RIU-1