The work aims to construct a classification and prediction model of medicine bottles. Differential Raman spectroscopy and X-ray fluorescence spectroscopy were used to analyze and test 54 plastic medicine bottles from different brands and producing areas. The differential Raman spectra of 54 samples and the content of chlorine, calcium, titanium, iron and zinc were obtained. The principal component analysis was used to reduce the dimension of the differential Raman spectrum data, and then the system clustering was used to divide the reduced dimension data into 8 categories. Based on this, the discriminant analysis model was established. The final discriminant model was verified by cross validation, and the accuracy was 90.7%, and the classification accuracy of the multi-layer perceptron was 100%. The classification effect was good. The differential Raman spectrum can infer the molecular structure of the sample according to the characteristic peaks in the spectrum, and the samples can be classified and analyzed according to the peak position. The X-ray fluorescence spectrum can distinguish the samples according to the types and contents of various elements, so as to realize the refinement in the group. Differential Raman spectroscopy and X-ray fluorescence spectroscopy can classify medicine bottles from organic and inorganic perspectives respectively. They complement each other in analysis. They can be used to explore a new angle and method of spectral combination for the actual handling of cases by public security organs.