Analysis and identification of glass products based on logistic regression

Abstract

Firstly, the correlation between categorical variables was studied, the relationship between non-parametric samples was established, the occurrence frequency of surface weathering and ornamentations, types and colors was counted, and the chi-square test model was established. Finally, the independence test was conducted. This time, for the analysis of the difference of multiple groups of data, the normal distribution test can be conducted on the chemical composition content proportion data, and the mean value fluctuation model can be established. Finally, through the significance test, it can be concluded that there is no obvious correlation between surface weathering and ornamentation, but ornamentation B is easier to be weathered than A and C. There is 99% certainty that "surface weathering is related to type", and high potassium is not easy to weathering, lead barium is easy to weathering; There was no significant correlation between surface weathering and color, but there were some differences between weathered and unweathered color species.