multivariate and univariate linear regression using MSE as cost function and gradient descent to minimize the cost function