Linear Regression

Linear Regression with one variable:-

In the problem where we predict share prices(regression problems), we have seen few parameters that are responsible for the price change. A regression problem considering a single parameter is called as Linear Regression with one variable.

Let's take export rate as our parameter and share price be our goal. A general plot be like,

We need a mathematical model to represent this trend. In this case, an equation of line touching maximum points.
Equation of line - y(o/p) = mx(i/p) + c
m = slope of line (responsible for the shape of line)
c = a constant that shifts the line vertically

In ML terms,
y becomes h(θ)
m becomes θ₁
c becomes θ၀

So our model is h(θ) = xθ₁ + θ၀

h(θ) is also known as hypothesis function.

For different values of θ₁ and θ၀ different lines will be obtained.

Our next task is to find the best model from these possibilities. Finding the best values for θ₁ and θ၀ is our goal. Learning Cost function will give us more idea about this.

On Next - Cost Function