It is a fact that Linear Regression has certain underlying conditions that need to be taken care of. However, one striking question that comes across my mind when I say this is; WHEN? In what context are these conditions relevant? For example, in as many as 3 books that dwell on Linear Regression (which I have read, of course there are more), either mention the conditions briefly or completely ignore those. Those books are; Think Stats (2nd edition) by Allen B. Downey, Data Science from Scratch by O' Reilly, and Introductory Statistics from OpenStax.
I was faced with this issue while working on a project for which my aim is to make predictions. Must I pay attention to the underlying conditions for Linear Regression? If yes, then which conditions weigh more than others? If I ignore those conditions and straightaway make predictions, what will be its statistical implications?
