Raw Raman spectral datasets of TEP, DIMP, and DMMP with a linear simulated dataset

The data support the study entitled "Prediction-space LOD calculation for nonlinear machine-learning Raman calibration: validation using RBF-SVR". The experimental Raman spectra were collected at different concentration levels to evaluate whether prediction-space definitions of quasi-blank prediction noise and low-concentration local sensitivity can be used for LOD calculation in nonlinear Raman calibration.

The linear simulated dataset was generated to test the consistency of the proposed LOD calculation strategy under near-linear conditions using PLSR as a reference model. Each file contains a Number column, a Concentration column, and spectral-variable columns corresponding to Raman shifts in cm^-1. The experimental datasets use percentage concentration labels, whereas the simulated dataset uses fractional concentration labels in scientific notation.