#!/usr/bin/python

def load_dataset():
    dataset = []
    with open('petrol.csv') as f:
        for line in f:
            sample = list(map(float, line.split(',')))
            x, y = sample[:-1], sample[-1]
            dataset.append((x, y))
    return dataset

def get_error(dataset, model):
    errors = []
    for x, y in dataset:
        delta = model(x) - y
        errors.append(delta * delta)
    return sum(errors) / len(errors)

def make_linear_model(weights):
    def model(x):
        y = sum(x[i] * weights[i] for i in range(len(x)))
        y += weights[-1]
        return y

    return model

dataset = load_dataset()
p = len(dataset[0][0])
parameters = [0 for _ in range(p + 1)]

rounds = 1000 * 1000
gamma = 10 ** -8

for i in range(rounds):
    model = make_linear_model(parameters)

    gradient = [0 for _ in range(p + 1)]
    for x, y in dataset:
        for j in range(p):
            gradient[j] += 2 * (x[j] * (model(x) - y)) / len(dataset)
        gradient[-1] += 2 * (model(x) - y) / len(dataset)

    if i % 10000 == 0 or i < 100:
        print("Iterace", i, ":")
        print("Parametry:", parameters)
        print("Střední kvadratická odchylka:", get_error(dataset, model))

        for x, y in dataset[:5]:
            print(x, model(x), y)
        print(gradient)

        print()

    for i in range(len(gradient)):
        parameters[i] -= gradient[i] * gamma

print("Výsledné parametry:", parameters)