https://medium.com/@monocosmo77/how-linear-function-approximation-is-used-in-reinforcement-learning-part1-machine-learning-94201f5459e3