import KNN
import matplotlib.pyplot as plt
from mpl_toolkits.mplot3d import Axes3D
g_Labels = ['largeDoses', 'smallDoses', 'didntLike']
g_color = ['red', 'orange', 'gray', 'blue']
if __name__ == '__main__':
print("\t\t\t============ Chap3 KNN ============")
x, y = DS.load_dataset(r'.\dataset.dat')
## convert y to label
y = np.array(list(map(lambda y1: g_Labels.index(y1), y)))
y = np.reshape(y, (-1, 1))
model = KNN.KNN_Model()
model.train(x, y)
x_test = DS.load_test_ds(r'.\testds.dat')
print(x_test)
y_test = model(x_test)
yy_test = list(g_Labels[int(y1)] for y1 in y_test)
print(yy_test)
'''draw plot'''
fig1 = plt.figure()
ax = Axes3D(fig1)
for i in range(len(y)):
_x = x[i]
_y = int(y[i])
X, y, _ = vector_label(trainpath)
X_test, y_test, testfilenames = vector_label(testpath)
print('训练集 %d个' % X.shape[0])
print('测试集 %d个' % X_test.shape[0])
# knn=KNN.KNN_Model(5)
# knn.fit(X,y)
# rand_indexs=random.sample([i for i in range(110)],10)
# for i in rand_indexs:
# p=knn.predict(np.array([X_test[i]]))
# print('文件%s 被预测为%d'%(testfilenames[i],p[0]))
# # 自己实现的
for k in [3, 5, 10]:
myknn = KNN.KNN_Model(k, metric='euclidean')
myknn.fit(X, y)
pred = myknn.predict(X_test)
print('k=%d 预测正确率为 %f' % (k, evaluate_acc(y_test, pred)))
print()
# sklearn
knn = KNeighborsClassifier(algorithm='brute')
# 参数选择
param_grid = {'n_neighbors': [i for i in range(3, 10)]}
grid = GridSearchCV(knn, param_grid, n_jobs=3)
grid.fit(X, y)
print("The best parameters are %s" % grid.best_params_)
model = grid.best_estimator_
y_pred = model.predict(X_test)
print('acc=%f' % accuracy_score(y_test, y_pred))
