A Line is Baseline of ML

I realize algebra is key to the way Machines learn. And the most basic component of algebra is a Line. In crudent understanding, a line is something that connects two points. But this simple concept when seen in context of regression, it becomes the base of prediction. The following algebric equation builds a line with dynamic slope, x-variable and y-intercept
y = mx + h
Now remember dependent variable, weight, independent variable and cost from machine learning.
The most efficient line built based on the available data can create the foundation of finding value of y (dependent variable) even for new values of x(independent variable) or put other way, you can tell size of watermelon or chicken, given number of days as input.
Add an activation function and suddenly non-linearity is added. Combined with backpropagation, its an explosion. We will discuss it some other day.
In the equation of line, x is independent variable. But what is independent variable? And how it impacts the learning is key question every AI enthusiast should ask himself.
With this background, my request to you is to put your thoughts on the nature and dimensions of a LINE.