I have 10 independent variable, how to identify the important variable for my Linear regression model ?

Question

Is p-value the only way to identify the important variable?

in progress 0
Xtramous 55 years 3 Answers 1040 views Contributor 0

Answers ( 3 )

  1. shankar
    1
    Reply

    Yes, the p-value is the first and most priority to check the significance of a variable. But there is another method like checking the correlation with target variable, add features, and check the adjusted r2 value whether it increases or decreases. If it increases then it is a significant variable else not.

  2. ravi_joe
    0
    Reply

    we can use p value although we can use feature selection techniques like rfecv(recursive feature elimination) or lasso regression can also be used for feature selection . we can also use scatter plots and corelation heatmap. finally pca may also be used but since the number of independent variables is less pca should not be used

  3. swap007
    0
    Reply

    1) Apart from checking individual p-values, you can also use step() function in R for
    feature selection. The step function builds many different models internally using different
    combinations of predictor variables and chooses the model with the least AIC.

    2) You can use Variable Importance Plot using decision trees which gives you the relative
    importance of all the features with respect to the most important one.

Leave an answer

Browse
Browse