SPSS data Analysis-least one multiplication

Linear regression is most commonly used as a fitting method with least squares, but the method is more susceptible to strong influence points, so when we fit the linear regression model, we also take the strong influence point as the condition to be considered. For strong-impact points, a more robust fit method is needed in cases where it cannot be corrected or deleted, and the least-squares method is the solution to such problems.

The least square method is due to the residual sum of squares, and the residual of the strong influence point is usually larger, after the square is larger, and the least one does not use the sum of squares and the sum of the absolute value, so for the residual of the strong impact point, its effect will be much smaller.

We compare the fitting effect of the least squares and the least squares when the strong impact points appear, and in SPSS, the least squares is the default fitting method for regression analysis, and the least-squares or other fitting methods need to be set manually.

First, make a scatter plot to get a preliminary judgment.

SPSS data Analysis-least one multiplication