It has two parts. One uses a self-developed perceptron training function. The other uses the perceptron class from Scikit-learn. For the first self-implementation, figures are output to show the progress of the training.