Quality prediction modeling of plastic extrusion process

Manufacturers are trying to improve their quality through predictive analysis using large amounts of data for smart manufacturing. However, it is difficult to select an appropriate quality prediction model since the characteristics of each process and data of companies are different. Therefore, in this study, the environmental data of plastic extrusion process from a manufacturing company were collected and analyzed by four models of logistic regression, support vector machine, random forest, and bagging method. The best model can be selected through performance evaluation using F1 score of each model. If the measurement data of the product are collected automatically in the future, a better method could be found.