In the statistical work of various industries, the relationship between dependent variables and independent variables is often studied and analyzed from different dimensions. For example, we need to count the traffic volume of Shanghai and Beijing in different time dimensions (morning, middle and evening). In this study, the data between them are not independent, and there is a certain relationship between them.
For the study of this kind of data, ordinary linear regression and logical regression cannot be used, and generalized estimation equation is needed to fit other models. Let's use SPSS software to show you how to make a statistical analysis of this situation. GEE is a quasi-likelihood estimation method based on generalized linear model, which is used to analyze data with intra-group correlation.
Working correlation matrix of generalized estimation equation;
1, etc., also known as commutative correlation. Rent key assumes that the correlation between any two observations is equal. This assumption is usually used for repeated measurement data that are not in chronological order.
2. Adjacent correlation, that is, only two adjacent observations are correlated.
3. autocorrelation, that is, correlation is related to the number of intervals, and the longer the interval, the smaller the correlation.
4. Uncertain correlation, that is, let the model estimate itself according to the data characteristics without specifying the correlation form in advance.
5. Independence, that is, irrelevance, assumes that the dependent variable is irrelevant.