import matplotlib.pyplot as plt
import numpy as np

from sklearn.datasets import make_moons
X, y = make_moons(noise=0.2, n_samples=200)

X.shape

y.shape

print(y)

plt.scatter(X[:, 0], X[:, 1], c=y, s=100)
plt.figsize(14,10)

from sklearn.cross_validation import train_test_split
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=.4)

X_train.shape

X_test.shape

from sklearn.neighbors import KNeighborsClassifier
knn = KNeighborsClassifier()
print(knn)


knn.fit(X_train, y_train)

y_predict = knn.predict(X_test)

plt.scatter(X_train[:, 0], X_train[:, 1], c=y_train, s=100)
plt.scatter(X_test[:, 0], X_test[:, 1], c=y_predict, alpha=0.3, s=100)
#plt.scatter(X_test[:, 0], X_test[:, 1], c='w', alpha=0.3, s=100)
plt.figsize(14,10)


y_predict

y_test

knn.score(X_test, y_test)

from sklearn.svm import LinearSVC
svm = LinearSVC()

svm.fit(X_train, y_train)

y_predict = svm.predict(X_test)

plt.figsize(14,10)
plt.scatter(X_train[:, 0], X_train[:, 1], c=y_train, s=100)
plt.scatter(X_test[:, 0], X_test[:, 1], c=y_predict, alpha=0.3, s=100)
w = svm.coef_.ravel()
a = -w[0] / w[1]
xx = np.linspace(-1, 2)
yy = a * xx - svm.intercept_ / w[1]
plt.plot(xx, yy)



import pandas as pd
train_data = pd.read_csv("kaggle_insult/train.csv")
test_data = pd.read_csv("kaggle_insult/test_with_solutions.csv")

y_train = np.array(train_data.Insult)
comments_train = np.array(train_data.Comment)
print(comments_train.shape)
print(y_train.shape)

comments_train[0], y_train[0]

comments_train[5], y_train[5]

from sklearn.feature_extraction.text import CountVectorizer
cv = CountVectorizer()
cv.fit(comments_train)
X_train = cv.transform(comments_train)

svm.fit(X_train, y_train)

comments_test = np.array(test_data.Comment)
y_test = np.array(test_data.Insult)
X_test = cv.transform(comments_test)
svm.score(X_test, y_test)

index = 8
comments_test[index], y_test[index], svm.predict(X_test.tocsr()[index])[0]