Note: You have tagged this question under ‘regression’. Hence I’ll stick to variable selection in regression.
It’s important to understand the variable importance in order to achieve high accuracy. The independent variables used in a model are meant to explain the maximum variance in dependent variable.
For example: We have a data set with 100 variable. We build a model using those 100 variables and get adjusted R2 as 80%. Adjusted R2 is nothing but explained variance in the dependent variable from independent variable.
Now, we did variable importance check and found that only 20 out of 100 are highly important. We build another model. This time we get adjusted R2 as 84%. This means,only 20 variables are sufficient enough to explain 84% variance in dependent variable.
I hope till here you’ve understand the concept of variable importance.
In regression, you can find variable importance in many ways. In linear regression, you can do forward selection, backward selection, stepwise selection etc. This helps in improving better and accurate models.
In logistic regression, you can check the significance of individual variables.
In SPSS, this metric is available in “Variables in the Equation Table”. In this table, focus on “Sig.” column which shows the significance of a variable in the model.
With 95% confidence interval, if Sig < 0.05, the variable will significant, else it won’t be significant. The image shown below will help you to understand it better.