Resolved: Linearity for Multiple Linear Regression
To examine this condition for a multiple linear regression, do we compare the individual variable/feature with the dependent variable?
Also, how similar is the concept of linearity with that of correlation?
Hi Jonathan!
Thanks for reaching out.
Yes, to examine linearity in multiple linear regression, you should check how each independent variable relates individually to the dependent variable. This can be done through scatter plots of each variable against the dependent variable, and by looking at residual plots after fitting the model. If the scatter and residual plots show a random pattern, this suggests a linear relationship.
As for the similarity between linearity and correlation - both concepts relate to the relationships between variables. But they focus on different aspects.
Linearity refers to a direct proportionate relationship that can be represented by a straight line in a model. This means that changes in one variable directly lead to proportional changes in another variable
Correlation measures how strongly two variables are related and move together. However, it does not specify that the relationship must be linear. For instance, there might be a strong correlation between the amount of time studied and test scores, but the increase in score per hour of study doesn’t have to be consistent across all ranges of study time.
Hope this helps.
Best,
Ivan
