Abstract: To evaluate the degree of unsaturation of different varieties of Camellia oil, it was necessary to establish a rapid evaluation method with a narrow iodine value range (iodine value difference less than 10). In this study, a rapid quantitative prediction model for iodine value of oil in high-resolution Raman spectroscopy based on linear regression and gradient descent method was established. The Raman signals (785 nm) about 39 group of Camellia oil samples and 10 group of commercially oils were firstly collected. Then, the intensity ratio of peaks of 1656 cm⁻¹ and 1440 cm⁻¹ (I_1656/1440) were selected through smoothing algorithm least squares smoothing filter (Savitzky-Golay), polynomial fitting and deconvolution algorithm Lorentzian. A credible model was obtained through correlation analysis with the iodine value of corresponding oil samples. The coefficient of determination (R²) of the test set of the constructed quantitative model was >0.82, the mean square error (MSE) was <0.73 and the root mean square error (RMSE) was <0.85. This quantitative model of edible oil iodine value can accurately and efficiently predict the unsaturation degree of Camellia oil, etc.