<p>Predicting the mixing ratio of molecules with similar structures from SERS spectra is expected to be applicable to disease diagnosis and environmental analysis. Currently, deep learning is applied to analyze SERS spectra due to the complexity of the spectrum shape. However, deep learning requires a lot of training data. In addition, deep learning models have complex structures, making it difficult to interpret the output results. In the field of analytical chemistry, it is often not possible to collect a large amount of measured data. Therefore, simple and interpretable machine learning models that require a small amount of training data are needed. This study investigates the multivariate analysis of surface-enhanced Raman scattering (SERS) spectra using simple machine learning models. The mixing ratios of the self-assembled monolayer (SAM) reagents of benzene thiol derivatives (as model molecules) were predicted from SERS spectra using several prediction models. Then, that prediction accuracy of each machine-learning model was compared. The results show that linear discriminant analysis (LDA) achieves the highest prediction accuracy (0.996). We analyze the SERS spectrum with a simple model such as LDA and achieve highly accurate predictions of the mixing ratios of the SAM reagents. This indicates that, even when the measurement data of the SERS spectra are scarce, highly accurate predictions can be made through analysis using a simple prediction model.</p>