This inquiry relates to gradient descent.

Gradient descent is a method that tries to iteratively discover the best parameters while minimizing the cost function. The mean squared error will be reduced in linear regression. The algorithm's general structure is as follows:

begin with some w=(w0,...,wn)

To minimize J(w), where J(w) is our cost function, keep altering w0,...,wn.

We will initially assign w0, w1,..., wn to all be zeros in this issue set. Then, we'll decide on a learning rate that will have an impact on how quickly w0,...,wn 0,..., changes. Finally, we will set N to the number of gradient descent iterations we desire to carry out.

Algorithm 1 defines the gradient descent pseudo-code for linear regression.

Write the function gradient_descent_one_variable(x, y, learning rate, n) that returns:

0 - a number representing the bias constant

1 - a number representing the weight constant

loss - a list that contains the MSE scores calculated during the gradient descent process.

The default value is 10^-5 for learning rate and 250 for n.

You can use mean_squared_error function for this task.

Algorithm 1: Gradient Descent for Linear Regression w0,w1,,wn0whileN=0dow0w0w0J(w)w1w1w1J(w)wnwnwnJ(w)w0w0w1w1wnwnlossJ(w)NN1endN