Variable Importance with Linear Regression Model

When we build Linear Regression models and look at the coefficient chart or table, we can’t help comparing the coefficients among the variables because at the end of the day we want to know which variables have greater effects on the target variable.

But we are not supposed do so because the units of each variable are different. For example, Order Quantity, Age, Weight, and Job Type are completely in different units and scales.

Of course, you can standardize the variables to make them have the ‘same’ scale, but there is a better way to address this challenge.

There is a method called ‘Relative Importance’, which can help you answer this very question of which variables have greater effects on the target variable.

And, you can use is simply by clicking on ‘Importance’ tab inside Linear Regression Analysis under Analytics View.

Basically, it gives a score to each variable based on how much it contributes to the overall R-Square.

There are a few methods you can choose and parameters inside the property.