KNN机器学习实战(sklearn)(包含训练数据)

本文链接：https://blog.youkuaiyun.com/weixin_42983055/article/details/82389392

这篇博客详细介绍了如何在sklearn库中应用KNN算法，通过实践得出n_neighbors为7或8时误差率最低，为0.002778。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

本文直接给出sklearn里面KNN 算法的用法。具体实现过程如下：


    # -*- coding: utf-8 -*-
    import numpy as np
    import pandas as pd
    from sklearn import datasets
    import operator
    from sklearn import neighbors
    import sklearn.model_selection as ms
    import matplotlib.pyplot as plt
    
    digits = datasets.load_digits()
    totalNum = len(digits.data)
    # 选出80%样本作为训练样本，其余20%测试
    trainNum = int(0.8 * totalNum)
    trainX,testX, trainY,testY = ms.train_test_split(digits.data, digits.target, random_state = 1, train_size = 0.8)
    
    np.shape(digits.data)
    np.shape(digits.target)
    
    #用图像来初步认识下特征的长相
    X_train = trainX.reshape(len(trainX), 8,8)
    X_train = X_train/X_train.max() # 数据归一化
    print("After reshaping, the shape of the X_train is:", X_train.shape)
    np.shape(X_train)
    a = X_train[1]
    a.shape
    plt.imshow(a, cmap = 'Greys_r')  #画图
    #训练模型，并计算不同K的情况下ER的变化情况
    ER = []
    for n_neighbors  in range(1,16):
        clf = neighbors.KNeighborsClassifier(n_neighbors, weights='uniform') #测试不同的K 对最终结果的影响
        clf.fit(trainX, trainY) #训练器
        Z = clf.predict(testX)  #预测
        x = 1- np.mean(Z == testY) #计算错误率
        ER.append(x) #将错误率储存在ER 中
    pd.DataFrame(ER).plot(title = 'the plot of error rate') #画图显示不同K对模型正确的影响

n_neighbors取值为1~15
通过以上的图形可知，n_neighbors = 7,8 时较为合适，此时的error rate 为0.002778

# -*- coding: utf-8 -*-
import numpy as np
from sklearn import neighbors, datasets
from sklearn.model_selection import train_test_split
from sklearn.utils.testing import assert_equal

rng = np.random.RandomState(0)
# load and shuffle digits
digits =