knn 鸢尾花实现，决策边界，k值，回归

#knn 鸢尾花实现，决策边界，k值，回归
import numpy as np
import matplotlib.pyplot as plt
# KNN分类器
from sklearn.neighbors import KNeighborsClassifier
#数据集
from sklearn import datasets
#数据集划分
from sklearn.model_selection import train_test_split

#鸢尾花数据导入
# iris是一个对象类型的数据
#其中包括了data（鸢尾花的特征）和target（分类标签）
#鸢尾花 标签(target)是0,1,2 
#标签名字(target_name)是setosa、versicolor,virginica
#瓣的长度、宽度、花萼的长度、宽度，一共4个特征
#shape(n,4)
iris = load_iris()
# 将样本与标签分开
x = iris[‘data‘] 
y = iris[‘target‘]
print(x.shape, y.shape)  # (150, 4) (150,)

#划分数据集，按照8:2分训练集与测试集
#通常3:1
x_train, x_test, y_train,y_test = train_test_split(
    x,y, test_size = 0.2)  # 8:2 
print(x_train.shape, x_test.shape, y_train.shape, y_test.shape) 
# (120, 4) (30, 4) (120,) (30,)

#使用kNeighborsClassifier训练模型
#参数k(n_neighbors)=5
#欧式距离(metric=minkowski & p=2)

clf = KNeighborsClassifier(n_neighbors=5, p=2, metric="minkowski")

#传递训练集
clf.fit(x_train, y_train)  # fit可以简单的认为是表格存储
# KNeighborsClassifier()

#评估模型，检验泛化性能
#predict函数进行模型预测，并计算出预测准确率
#返回预测标签
#y_predict 用x_test测试，分类
#返回预测的x_test标签属于哪类
#然后验证预测的标签和结果有多少一样，计算预测准确率
y_predict = clf.predict(x_test)
y_predict.shape  # (30,)
acc = sum(y_predict == y_test) / y_test.shape[0]
acc

#不预测x_test标签属于哪类
#直接用score获取预测准确率
print(" {:.3f}".format(clf.score(x_test, y_test)))

#预测新数据标签
x_new = np.array([[4, 2.5, 1.6, 0.3]])
prediction = clf.predict(x_new)

print("鸢尾花种类: {}".format(
    iris[‘target_names‘][prediction]))

#k取不同值时，训练集准确率和测试集准确率
#k取不同值时，记录2个准确率，然后x取k值，y取准确度，画2个折线图

from sklearn import neighbors
from sklearn.model_selection import train_test_split
from sklearn.datasets import load_iris
from matplotlib import pyplot as plt 
plt.rcParams[‘font.sans-serif‘] = [‘Arial Unicode MS‘]
from pylab import *
#鸢尾花按照8:2分训练集与测试集
iris = load_iris()
x = iris[‘data‘] 
y = iris[‘target‘]
x_train, x_test, y_train,y_test = train_test_split(
    x,y, test_size = 0.2)  # 8:2 


#分别保存在不同的邻居个数前提下
#模型在训练集与测试集上的预测准确率
training_accuracy = []  #训练集预测准确率
test_accuracy = []      #测试集预测准确率

# n_neighbors取值：1,3,5,7,9,11
n = [1,3,5,7,9,11]

for n_neighbors in n:
    # 构建模型
    clf = neighbors.KNeighborsClassifier(n_neighbors=n_neighbors)
    clf.fit(x_train, y_train)

    # 记录训练集预测准确率
    training_accuracy.append(clf.score(x_train, y_train))
    # 记录测试（泛化）集预测准确率
    test_accuracy.append(clf.score(x_test, y_test))


#对结果进行可视化展示    
plt.plot(n, training_accuracy, 
         label=U"训练精度值")
plt.plot(n, test_accuracy, 
         label=U"测试精度值")

plt.ylabel(U"预测精度值")
plt.xlabel(U"邻居个数")
plt.legend()

plt.show()

#决策边界（划分范围）
#需要等高线基础
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
from sklearn.model_selection import train_test_split
from sklearn.metrics import accuracy_score
#导入iris数据
from sklearn.datasets import load_iris
iris = load_iris()
X=iris.data
y=iris.target
X=X[:,:2] #只取前两列

X_train, X_test, y_train, y_test = train_test_split(
X, y, stratify=y, random_state=42) #划分数据，random_state固定划分方式
#导入模型
from sklearn.neighbors import KNeighborsClassifier 
#训练模型
knn = KNeighborsClassifier(n_neighbors=5)
knn.fit(X_train, y_train)
y_pred = knn.predict(X_test)


def plot_decision_boundary(clf , axes):
    xp=np.linspace(axes[0], axes[1], 300) #均匀300个横坐标
    yp=np.linspace(axes[2], axes[3], 300) #均匀300个纵坐标
    x1, y1=np.meshgrid(xp, yp) #生成300x300个点
     #np.c_:是按行连接两个矩阵，就是把两矩阵左右叠加，要求行数相等
    xy=np.c_[x1.ravel(), y1.ravel()] 
    y_pred = clf.predict(xy).reshape(x1.shape) #训练之后平铺
    custom_cmap = ListedColormap([‘#fafab0‘,‘#9898ff‘,‘#a0faa0‘])
    plt.contourf(x1, y1, y_pred, alpha=0.3, cmap=custom_cmap)

    
plot_decision_boundary(knn, axes=[4, 8, 1.5, 5])

#画三种类型的点
p1=plt.scatter(X[y==0,0], X[y==0, 1], color=‘blue‘)
p2=plt.scatter(X[y==1,0], X[y==1, 1], color=‘green‘)
p3=plt.scatter(X[y==2,0], X[y==2, 1], color=‘red‘)
#设置注释
plt.legend([p1, p2, p3], iris[‘target_names‘], loc=‘upper right‘)
plt.show()

#https://www.cnblogs.com/onenoteone/p/12441726.html

#回归
# 生成样本数据
np.random.seed(0)
X = np.sort(5 * np.random.rand(40, 1), axis=0)
T = np.linspace(0, 5, 500)[:, np.newaxis]
y = np.sin(X).ravel()
# 给目标集添加一些噪音数据
y[::5] += 1 * (0.5 - np.random.rand(8))
# 训练回归预测模型,k值分别去1,3,5,7,9,11
for i, n_neighbors in enumerate([1,3,5]):
    knn = neighbors.KNeighborsRegressor(n_neighbors=n_neighbors)
    y_ = knn.fit(X, y).predict(T)
    plt.subplot(3, 1, i + 1)
    plt.scatter(X, y, color=‘darkorange‘, label=‘data‘)
    plt.plot(T, y_, color=‘navy‘, label=‘prediction‘)
    plt.axis(‘tight‘)
    plt.legend()
    plt.title("KNeighborsRegressor (k = %i,R^2 = %.3f)" % 
              (n_neighbors,knn.score(X, y)))
plt.tight_layout()

plt.show()