可能是上手最合適的Knn項目

KNN的中文，有標(biāo)簽（監(jiān)督），樣本數(shù)量小于100K的非文本數(shù)據(jù)（位置相近，不是邏輯相近，離散）

近朱者赤近墨者黑的意思

from sklearn import datasets
from sklearn.neighbors import KNeighborsClassifier
from sklearn.model_selection import train_test_split
# import matplotlib.pyplot as plt
from pylab import *
"""
http://scikit-learn.org/dev/modules/generated/sklearn.datasets.load_iris.html
"""


iris = datasets.load_iris()  # 3類*50樣本*4特征 鳶yuan尾屬植物
# 總數(shù)據(jù)的0.5作為訓(xùn)練數(shù)據(jù) 另外0.5作為測試數(shù)據(jù)
train_data, test_data, train_label, test_label = train_test_split(iris.data, iris.target, train_size=0.5)
# 初始化設(shè)定KNN分類器，括號里面可以使用參數(shù)，可以看函數(shù)定義或者官網(wǎng)n_neighbors weights algorithm 等這些參數(shù)可以定義
knn = KNeighborsClassifier()
knn.fit(train_data, train_label)  # sklearn經(jīng)典命令~~ 向分類器中喂數(shù)據(jù)
prediction = knn.predict(test_data)  # 預(yù)測
# 打印預(yù)測結(jié)果
print('預(yù)測值', prediction)
print('實際值', test_label)
score = len(prediction)
for i in range(len(prediction)):
    if prediction[i] != test_label[i]:
        score -= 1
print('正確率', score/len(prediction), '錯誤個數(shù)', len(prediction)-score)


# 查看數(shù)據(jù)集，和預(yù)測沒有關(guān)系
x_index = 0
y_index = 3
mpl.rcParams['font.sans-serif'] = ['SimHei']
colors = ['red', 'orange', 'green']
names_sc = ['清風(fēng)藤', '云芝', '錦葵']
feature_names_sc = ['萼片長度', '萼片寬度', '花瓣長度', '花瓣寬度']
# setosa清風(fēng)藤 versicolor云芝 virginica錦葵
for label, color in zip(range(len(names_sc)), colors):
    plt.scatter(iris.data[iris.target == label, x_index],
                iris.data[iris.target == label, y_index],
                label=names_sc[label],
                c=color)
plt.xlabel(feature_names_sc[x_index])
plt.ylabel(feature_names_sc[y_index])
plt.legend(loc='upper left')
plt.show()