The digitization of injection molding processes is being developed to improve productivity without depending on the know-how of skilled workers. For example, in-mold sensors have been used to shorten production preparation time by correcting the machine difference after the production base was moved to a foreign country. The correlation between the number of sensor features, the molding conditions, and the quality of the molded product needs to be analyzed to optimize the molding conditions with these in-mold sensors. In this study, the method for correcting the machine difference in the injection molding machine was investigated using data analyzed by in-mold sensors. Injection molding was performed by setting the same molding conditions across multiple machines with different screw diameters. The number of sensor features correlated with the temperature, velocity, and pressure were found to vary across machines. As a result of the regression analysis between the molding conditions and the number of sensor features correlated with these parameters, the setting and measured values were found to fit well using the linear regression analysis model. The error rate of the molded product weight between two different molding machines was improved using the obtained model, and the molding conditions could be predicted when matching the number of sensor features.