2) Since the updating of learning is follow the rule:
w = w + learning_rate * (expected - predicted) * x
with x equal to zero we would get same value of w no matter which learning rate we would take
3) https://colab.research.google.com/drive/18wEXqxYov1At0Jriuh53vBWc2z-aBHNV?usp=sharing
I implement the perceptron according to the article found on medium, and tried to implement more data. That gave me proof to poor usage of perceptron as a learning method.