Aiming at the problems of conventional linear regression and traditional neural network models in predicting the quality of electrolytic copper, such as poor fitting effect and low prediction accuracy, the quality of machine learning algorithms based on the maximum information coefficient, Random Forest algorithm, support vector machine, and correlation vector machine is established. The prediction model is used to reveal the non-linear fluctuation law of electrolytic copper quality under the influence of multiple factors. Taking the electrolytic copper production data of a copper industry company as an example, based on 19 related factors affecting the quality of electrolytic copper and 5 electrolytic copper quality control index systems, the maximum information coefficient is first used to analyze the non-linear correlation between the influencing factors. Secondly, the Random Forest algorithm is used to clarify the main control factors that affect the quality of electrolytic copper. Finally, the particle swarm optimization support vector machine and correlation vector machine are used to establish the quality prediction model of electrolytic copper. The results show that the RF algorithm is used to characterize the complex nonlinear relationship between the quality of electrolytic copper and the quality of electrolytic copper, and the five main controlling factors of electrolytic copper are clarified, which is confirmed by the nonlinear correlation analysis based on MIC. For all 19 influencing factors, the RVM model is very similar in prediction accuracy to the PSO-LSSVM model (the gaps are 4.45% and 14.16%, respectively), and are better than the conventional multiple linear regression model and the traditional neural network model. Quality prediction model of electrolytic copper based on RF-RVM model, the prediction error of the test data set is smaller than that of the RVM model (the lowest error index value is less than 5%). The research results have certain reference value for electrolytic copper quality control.