Các bước kiểm tra các giả định và đơn giản hóa mô hình trong các mô hình hồi quy tuyến tính nhiều biến

Abstract

The multiple linear regression models are widely applied in various fields such as economics, medicine, social sciences, and engineering. In the multiple linear regression models, ensuring the validity of model assumptions and simplifying the model are crucial steps to obtain accurate and reliable statistical inferences. The main purpose of the paper is to systematically list and scientifically test the assumptions of the model. By examining assumptions regarding linearity, homoscedasticity, normality of model error distribution, and multicollinearity, we can ensure that the multiple linear regression model can be applied to data. Additionally, the paper focuses on exploring methods to simplify the model and test hypotheses for sub-models, highlighting the utility of these methods in creating simpler models while maintaining the accuracy of model predictions. The methods for testing assumptions and simplifying the model are applied to the abrasion resistance of rubber data. Applying these methods to the rubber abrasion resistance data helps optimize the analysis process and provides a better understanding of the factors influencing the abrasion resistance of this material.