import array
import math
import random
import matplotlib.pyplot as plt
import numpy as np
from matplotlib.colors import ListedColormap
from sklearn.model_selection import train_test_split
from sklearn.datasets import make_moons, make_circles, make_classification
from sklearn.neural_network import MLPClassifier
from sklearn.linear_model import LinearRegression, Lasso, Ridge, Perceptron
from sklearn.metrics import accuracy_score
from flask import Flask
app = Flask(__name__)
@app.route("/")
def home():
return "" \
"Жукова Алина ПИбд-41
" \
"Лабораторная работа №1
" \
"" \
""
# Работа с типовыми наборами данных и различными моделями
# сгенерируйте определенный тип данных и сравните на нем 3 модели
# 10.Данные: make_moons (noise=0.3, random_state=rs)
# Модели:
# · Линейную регрессию
# · Многослойный персептрон с 10-ю нейронами в скрытом слое (alpha = 0.01)
# · Многослойный персептрон со 100-а нейронами в скрытом слое (alpha = 0.01)
@app.route("/k4_1_task_1", methods=['GET'])
def k4_1_task_1():
X, y = make_classification(n_features=2, n_redundant=0, n_informative=2,
random_state=0, n_clusters_per_class=1)
rng = np.random.RandomState(2)
X += 2 + rng.uniform(size=X.shape)
linearly_dataset = (X, y)
cm_bright = ListedColormap(['#FF0000', '#0000FF'])
cm_bright2 = ListedColormap(['#FF000066', '#0000FF66'])
moon_dataset = make_moons(noise=0.3, random_state=0)
circles_dataset = make_circles(noise=0.2, factor=0.5, random_state=1)
datasets = [moon_dataset, circles_dataset, linearly_dataset]
X, y = moon_dataset
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=.2, random_state=42)
ridge_regression = Ridge(alpha=3, random_state=240)
ridge_regression.fit(X_train, y_train)
linear_accuracy = str(ridge_regression.score(X_test, y_test))
plt.subplot(1, 3, 1)
plt.scatter(X_train[:, 0], X_train[:, 1], c=y_train, cmap=cm_bright)
plt.scatter(X_test[:, 0], X_test[:, 1], c=y_test, cmap=cm_bright2)
x_min = moon_dataset[0][:, 0].min()
g_min = None
y_min = None
x_max = moon_dataset[0][:, 0].max()
g_max = None
y_max = None
for k in range(-21, 50):
elem = np.array([[x_min, k / 20]])
getted = ridge_regression.predict(elem)
if (g_min == None or math.fabs(0.5 - getted) < math.fabs(0.5 - g_min)):
g_min = getted
y_min = elem[0][1]
else:
if(math.fabs(0.5 - getted) > math.fabs(0.5 - g_min)):
break
for k in range(-21, 50):
elem = np.array([[x_max, k / 20]])
getted = ridge_regression.predict(elem)
if (g_max == None or math.fabs(0.5 - getted) < math.fabs(0.5 - g_max)):
g_max = getted
y_max = elem[0][1]
else:
if(math.fabs(0.5 - getted) > math.fabs(0.5 - g_max)):
break
x = ridge_regression.predict(X_test)
plt.plot([x_min, x_max], [y_min, y_max], label="line", color="yellow")
# plt.show()
# Перцептрон 10 скрытых слоев
perceptr = MLPClassifier(random_state=1, max_iter=2000, n_iter_no_change=20, activation="tanh",
alpha=0.01, hidden_layer_sizes=[10,], tol=0.00000001)
perceptr.fit(X_train, y_train)
prediction = perceptr.predict(X_test)
perceptron_accuracy = str(accuracy_score(y_test, prediction))
prediction = perceptr.predict(moon_dataset[0])
perceptron_accuracy_all = str(accuracy_score(moon_dataset[1], prediction))
params_set = []
y_elem = None
g_elem = None
for data_elem in moon_dataset[0]:
for k in range(-21, 50):
elem = np.array([[data_elem[0], k / 20]])
getted = perceptr.predict(elem)
if (g_elem == None and getted == 0):
params_set.append([data_elem[0], -21 / 20])
g_elem = None
else:
if(getted == 1 and (getted == g_elem or g_elem == None)):
g_elem = getted
y_elem = elem[0][1]
else:
params_set.append([data_elem[0], y_elem])
g_elem = None
break
if (g_elem != None):
params_set.append([data_elem[0], 50 / 20])
g_elem = None
params_set.sort()
params_set = np.array(params_set)
plt.subplot(1, 3, 2)
plt.scatter(X_train[:, 0], X_train[:, 1], c=y_train, cmap=cm_bright)
plt.scatter(X_test[:, 0], X_test[:, 1], c=y_test, cmap=cm_bright2)
plt.plot(params_set[:, 0], params_set[:, 1], label="line", color="yellow")
# plt.show()
# Перцептрон 100 скрытых слоев
perceptr100 = MLPClassifier(random_state=1, max_iter=2000, n_iter_no_change=20, activation="tanh",
alpha=0.01, hidden_layer_sizes=[100, ], tol=0.00000001)
perceptr100.fit(X_train, y_train)
prediction = perceptr100.predict(X_test)
perceptron100_accuracy = str(accuracy_score(y_test, prediction))
prediction = perceptr100.predict(moon_dataset[0])
perceptron100_accuracy_all = str(accuracy_score(moon_dataset[1], prediction))
params_set = []
y_elem = None
g_elem = None
for data_elem in moon_dataset[0]:
for k in range(-21, 30):
elem = np.array([[data_elem[0], k / 20]])
getted = perceptr100.predict(elem)
if (g_elem == None and getted == 0):
params_set.append([data_elem[0], -21 / 20])
g_elem = None
else:
if(getted == 1 and (getted == g_elem or g_elem == None)):
g_elem = getted
y_elem = elem[0][1]
else:
params_set.append([data_elem[0], y_elem])
g_elem = None
break
if (g_elem != None):
params_set.append([data_elem[0], 30 / 20])
g_elem = None
params_set.sort()
params_set = np.array(params_set)
plt.subplot(1, 3, 3)
plt.scatter(X_train[:, 0], X_train[:, 1], c=y_train, cmap=cm_bright)
plt.scatter(X_test[:, 0], X_test[:, 1], c=y_test, cmap=cm_bright2)
plt.plot(params_set[:, 0], params_set[:, 1], label="line", color="yellow")
plt.show()
return "" \
"Работа с типовыми наборами данных и различными моделями
" \
"Вариант 10. Данные: make_moons (noise=0.3, random_state=rs)
" \
"Модели:\n 1) Линейная регрессия" \
"\n 2) Многослойный персептрон с 10-ю нейронами в скрытом слое (alpha = 0.01)" \
"\n 3) Многослойный персептрон со 100-а нейронами в скрытом слое (alpha = 0.01)
" \
"Оценка точности линейной регрессии: " + linear_accuracy + "
" \
"Оценка точности (тестовые данные) перцептрона 10 нейронов в скрытом слое: " + perceptron_accuracy + "
" \
"Оценка точности (тестовые данные) перцептрона 100 нейронов в скрытом слое: " + perceptron100_accuracy + "
" \
"Оценка точности (все точки) перцептрона 10 нейронов в скрытом слое: " + perceptron_accuracy_all + "
" \
"Оценка точности (все точки) перцептрона 100 нейронов в скрытом слое: " + perceptron100_accuracy_all + "
" \
""
if __name__ == "__main__":
app.run(debug=True)