So today is day 1 of 100.
Built linear regression from scratch in NumPy. Intuition, MSE, gradient descent by hand first, then code, then checked against scikit-learn.
Small loop: predict, check the error, adjust, repeat. A lot of ML is built on that.
