from sklearn import datasets, linear_model
from sklearn.metrics import r2_score, mean_squared_error
from sklearn.preprocessing import PolynomialFeatures
from sklearn.linear_model import LinearRegression
from sklearn.pipeline import Pipeline
from sklearn.model_selection import cross_val_score
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt

## подкл выборку

x_train = pd.read_table ("tic/ticdata2000.txt", header = None).iloc [0:4000, 0:85]
x_test = x_train.iloc[0:4000, 84]


y_train = pd.read_table ("tic/ticeval2000.txt", header = None)
y_test = pd.read_table ("tic/tictgts2000.txt", header = None)


## обучение выборки
regression = linear_model.LinearRegression()
regression.fit (x_train, y_train)


train_predict = regression.predict (x_train)

print('Коэффициенты: \n', regression.coef_)
print("Cреднеквадратичная ошибка: %.2f" % mean_squared_error(y_train, train_predict))
print('Оценка отклонения: %.2f' % r2_score(y_train, train_predict))

## Проверка точности модели по тестовой выборке и запись в результирующий файл
y_pr = pd.DataFrame(train_predict)
y_test = y_test.reset_index(drop = True)
res = pd.concat([y_pr, y_test], axis=1)
res.to_csv("result.txt", index = False)

## Построение модели с использованием полиномиальной функции (где-то тут ошибка)

degrees = [1,2,3]
err = [1,2,3]

pol_index = x_test
x_train_pol = x_train
y_train_pol = y_train
x_test_pol = x_test
y_test_pol = y_test

for i in range(len(degrees)):
    polynomial_features = PolynomialFeatures(degree = degrees[i],
                                             include_bias = False)
    linear_regression = LinearRegression()
    pipeline = Pipeline([("polynomial_features", polynomial_features),
                         ("linear_regression", linear_regression)])
    pipeline.fit(x_train_pol, y_train_pol)
    scores = cross_val_score(pipeline, x_train_pol, y_train_pol,
                             scoring="neg_mean_squared_error")
    pred_y = pipeline.predict(x_test_pol)
    err[i] = -scores.mean()
    print("\nСтепень: {}\nСреднеквадратичная ошибка = {}(+/- {})".format(degrees[i], -scores.mean(),
                                               scores.std()))
    print('Показатель отклонения: %.3f' % r2_score(y_test_pol, pred_y))