Aiming at the identification of adulterated olive oil, based on the fusion of near-infrared and Raman data on the feature level, a model for the identification of adulterated olive oil was established.24 pure olive oil samples and 248 adulterated olive oil samples were used for model establishment in the experiment. With three parameter optimization methods, support vector machine classification and regression models were established based on fusion of near-infrared and Raman data on the feature level with canonical correlation analysis (CCA). The results showed that the accuracy of correction and prediction sets of olive oil adulteration classification model based on the fusion of near-infrared and Raman data was high, and the correlation coefficient of prediction model based on the fusion of near-infrared and Raman data reached 99.13%. The results showed that the classification and prediction models were feasible and effective in the identification of adulterated olive oils, and this work shed light on the development and applications of fusion of spectral data.