SPSS data analysis-Simple linear regression

Regression analysis can also describe the relationship between the two variables, but they also differ, and the correlation analysis can describe the degree of tightness between the variables by the correlation coefficient size, and the regression analyses can not only describe the tightness between the variables, but also quantitatively describe when a variable changes, The degree of influence on another variable is not possible by correlation analysis, and because of this, regression analyses are more used to predict and control variable values, but regression analysis is not the same as causality.

According to the different models can be divided into linear regression and nonlinear regression

Linear regression analysis is generally described by the linear model, and the variance analysis model, just the different parts of the term, the regression model is divided into constants, regression parts, residuals
Constant is the so-called intercept, the regression part by the regression coefficient and the independent variable, here the regression coefficient can also be a slope, residual is the difference between the predicted value and the measured value, meaning for all the random and non-random factors can not be estimated by the independent variables caused by the variation. These concepts are similar to the variance analysis model.

Linear regression also has certain applicable conditions
1. Linear trend, that is, between the independent variable and the dependent variable is linear, this is the most basic requirement of linear regression, if not, it can not be analyzed with linear regression, this is possible by the scatter plot to confirm
2. Independence, that is, the value of the dependent variable is independent of each other, reflected in the model, is to require the residual difference between independent
3. Normality, that is, requires the dependent variable to obey the normal distribution, reflected in the model, is to require residual residuals to obey the normal distribution
4. Variance homogeneity, that is, requires that the dependent variable has the same variance, reflected in the model, that is, the residual error is required to have the same variance
5. There can be no high correlation between independent variables

Above 3, 4 points will affect the predicted value and prediction interval of the dependent variable, and the 5th should be paid special attention in multiple linear regression of several independent variables.

A linear regression of only one argument and one dependent variable is called simple linear regression, and this section mainly introduces simple linear regression.

ANALYSIS-Regression-linear

SPSS data analysis-Simple linear regression