When building a model, a critical step is having an accurate baseline. If you get 95% of the accuracy doing linear regression, it’s not worth optimizing for that last five percent.
But sometimes it can be. Discuss this beforehand so everyone agrees on the subject