机器学习 | KNN, K近邻算法

# 从sklearn.neighbors里导入 KNN分类器的类
from sklearn.neighbors import KNeighborsClassifier
# 通过类实例化一个knn分类器对象
# 类中的具体参数
# KNeighborsClassifier(n_neighbors=5,weights='uniform',algorithm='auto',leaf_size=30,p=2, metric='minkowski',metric_params=None,n_jobs=None,**kwargs,）
knn_clf = KNeighborsClassifier(n_neighbors=k）
# 通过对象调fit(）方法, 传入训练集， 训练模型
knn_clf.fit(X_train, y_train）X
# 训练好的模型， 通过其他接口， 传入测试集查看模型效果
knn_clf.score(X_test, y_test）
# 预测结果
knn_clf.predict(X_test)
knn_clf.predict_proba(X_test)

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import matplotlib as mpl
from sklearn.model_selection import train_test_split
from sklearn.metrics import accuracy_score
from matplotlib.colors import ListedColormap
#导入iris数据
from sklearn.datasets import load_iris
iris = load_iris()
X=iris.data[:,:2] #只取前两列
y=iris.target

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
n_neighbors = 5
knn = KNeighborsClassifier(n_neighbors=n_neighbors)
knn.fit(X_train, y_train)
y_pred = knn.predict(X_test)
#查看各项得分
print("y_pred",y_pred)
print("y_test",y_test)
print("score on train set", knn.score(X_train, y_train))
print("score on test set", knn.score(X_test, y_test))
print("accuracy score", accuracy_score(y_test, y_pred))

# 可视化

# 自定义colormap
def colormap():
    return mpl.colors.LinearSegmentedColormap.from_list('cmap', ['#FFC0CB','#00BFFF', '#1E90FF'], 256)

x_min, x_max = X[:, 0].min() - 1, X[:, 0].max() + 1
y_min, y_max = X[:, 1].min() - 1, X[:, 1].max() + 1
axes=[x_min, x_max, y_min, y_max]
xp=np.linspace(axes[0], axes[1], 500) #均匀500的横坐标
yp=np.linspace(axes[2], axes[3],500) #均匀500个纵坐标
xx, yy=np.meshgrid(xp, yp) #生成500X500网格点
xy=np.c_[xx.ravel(), yy.ravel()] #按行拼接，规范成坐标点的格式
y_pred = knn.predict(xy).reshape(xx.shape) #训练之后平铺

# 可视化方法一
plt.figure(figsize=(15,5),dpi=100)
plt.subplot(1,2,1)
plt.contourf(xx, yy, y_pred, alpha=0.3, cmap=colormap())
#画三种类型的点
p1=plt.scatter(X[y==0,0], X[y==0, 1], color='blue',marker='^')
p2=plt.scatter(X[y==1,0], X[y==1, 1], color='green', marker='o')
p3=plt.scatter(X[y==2,0], X[y==2, 1], color='red',marker='*')
#设置注释
plt.legend([p1, p2, p3], iris['target_names'], loc='upper right',fontsize='large')
#设置标题
plt.title(f"3-Class classification (k = {n_neighbors})", fontdict={'fontsize':15} )

# 可视化方法二
plt.subplot(1,2,2)
cmap_light = ListedColormap(['pink', 'cyan', 'cornflowerblue'])
cmap_bold = ListedColormap(['darkorange', 'c', 'darkblue'])
plt.pcolormesh(xx, yy, y_pred, cmap=cmap_light)

# Plot also the training points
plt.scatter(X[:, 0], X[:, 1], c=y, cmap=cmap_bold,
                edgecolor='k', s=20)
plt.xlim(xx.min(), xx.max())
plt.ylim(yy.min(), yy.max())
plt.title(f"3-Class classification (k = {n_neighbors})" ,fontdict={'fontsize':15})
plt.show()

# 导入乳腺癌数据集的类及其他包
from sklearn.datasets import load_breast_cancer
from sklearn.neighbors import KNeighborsClassifier
from sklearn.model_selection import train_test_split
# 实例化一份乳腺癌数据集对象
breast_cancer= load_breast_cancer()
# 查看数据
breast_cancer
# 数据的特征, 返回一个二维数组
X = breast_cancer['data']
X = pd.DataFrame(X)
name = ['平均半径','平均纹理','平均周长','平均面积','平均光滑度','平均紧凑度','平均凹度','平均凹点','平均对称','平均分形维数','半径误差','纹理误差','周长误差','面积误差','平滑度误差','紧凑度误差','凹度误差','凹点误差','对称误差','分形维数误差','最差半径','最差纹理','最差的边界','最差的区域','最差的平滑度','最差的紧凑性','最差的凹陷','最差的凹点','最差的对称性','最差的分形维数']
X.columns = name

# 数据的标签， 返回一个一维数组
y = breast_cancer['target']
# 划分数据，random_state固定划分方式
X_train, X_test, y_train, y_test = train_test_split(X, y, stratify=y,random_state=42) 

# 实例化一个5个最近邻的knn分类器
knn_clf = KNeighborsClassifier(n_neighbors=5)
# 训练模型
knn_clf.fit(X_train, y_train)
# 测试模型的准确率
knn_clf.score(X_test, y_test)
# 0.9590643274853801

# 导入归一化类
from sklearn.preprocessing import MinMaxScaler
mms = MinMaxScaler() # 实例化对象
mms.fit(X_train) # 这一步是在学习训练集，生成训练集上的极小值和极差
X_train = mms.transform(X_train) # 用训练集上的极小值和极差归一化训练集
X_test = mms.transform(X_test) # 用训练集上的极小值和极差归一化测试集

import numpy as np  
from sklearn.cross_validation import StratifiedKFold  
kfold = StratifiedKFold(y=y_train,   
                        n_folds=10,  
                        random_state=1)  
scores = []  
for k, (train, test) in enumerate(kfold):  
    model.fit(X_train[train], y_train[train])  
    score = model.score(X_train[test], y_train[test])  
    scores.append(score)
print('\nCV accuracy: %.3f +/- %.3f' % (np.mean(scores), np.std(scores)))

from sklearn.model_selection import ShuffleSplit
n_samples = iris.data.shape[0]
cv = ShuffleSplit(n_splits=5, test_size=0.3, random_state=0)
cross_val_score(clf, iris.data, iris.target, cv=cv)  
array([0.977..., 0.977..., 1.  ..., 0.955..., 1.])

from sklearn.model_selection import cross_val_score as cvs
L = []
for k in range(1, 21):
 knn_clf = KNeighborsClassifier(n_neighbors=k)
 result = cvs(knn_clf,X_train, # 使用训练集
               y_train,
              cv = 5) # 5折交叉验证
    result_mean = result.mean()
    result_var = result.var()
    L.append((k, result_mean, result_var))
score = pd.DataFrame(L)
score.columns = ['k', '平均准确率', '方差']
score

import matplotlib.pyplot as plt
plt.style.use('seaborn')
plt.rcParams['font.sans-serif']=['Simhei'] #显示中文 
plt.rcParams['axes.unicode_minus']=False  #显示负号
plt.figure(figsize=(8, 6), dpi=100)
plt.plot(score.k, score.平均准确率)
# 再画两条偏离均值2倍方差的曲线
plt.plot(score.k, score.平均准确率+2*score.方差, linestyle='**', color='r')
plt.plot(score.k, score.平均准确率-2*score.方差, linestyle='--', color='r')
plt.xticks(score.k)
plt.xlabel('k值', fontsize=20)
plt.ylabel('平均准确率', fontsize=20)
plt.title("k值学习曲线")
# plt.savefig('learning_curve_picture.jpeg')

import numpy as np
import matplotlib.pyplot as plt
from matplotlib.colors import ListedColormap
from sklearn import neighbors, datasets

n_neighbors = 10
h = .02  
breast_cancer= datasets.load_breast_cancer()
X = breast_cancer['data'][:, :2]
y = breast_cancer['target']

cmap_light = ListedColormap(['orange', 'cornflowerblue'])
cmap_bold = ListedColormap([ 'darkblue', 'darkorange'])

p1=plt.figure(figsize=(15,5),dpi=100) 
num = 1
for weights in ['uniform', 'distance']:
    ax1=p1.add_subplot(1,2,num)
    num+=1
    clf = neighbors.KNeighborsClassifier(n_neighbors, weights=weights)
    clf.fit(X, y)
    x_min, x_max = X[:, 0].min() - 1, X[:, 0].max() + 1
    y_min, y_max = X[:, 1].min() - 1, X[:, 1].max() + 1
    xx, yy = np.meshgrid(np.arange(x_min, x_max, h),
                         np.arange(y_min, y_max, h))
    Z = clf.predict(np.c_[xx.ravel(), yy.ravel()])
    Z = Z.reshape(xx.shape)
    plt.pcolormesh(xx, yy, Z, cmap=cmap_light)
    plt.scatter(X[:, 0], X[:, 1], c=y, cmap=cmap_bold,
                edgecolor='k', s=20)
    plt.xlim(xx.min(), xx.max())
    plt.ylim(yy.min(), yy.max())
    plt.title(" (k = %i, weights = '%s')"% (n_neighbors, weights))
plt.show()
plt.savefig('weight.tiff')