一个简单的分类模型knn

#导入数据
1 %matplotlib notebook
import numpy as np
import matplotlib.pyplot as plt
import pandas as pd
from sklearn.model_selection import train_test_split

fruits = pd.read_table('fruit_data_with_colors.txt')

fruits.head()

# 创建从fruit_label到fruit_name的映射
lookup_fruit_name = dict(zip(fruits.fruit_label.unique(), fruits.fruit_name.unique()))   
lookup_fruit_name

{1: 'apple', 2: 'mandarin', 3: 'orange', 4: 'lemon'}

检验数据

# 画点阵
from matplotlib import cm

X = fruits[['height', 'width', 'mass', 'color_score']]
y = fruits['fruit_label']
#划分train和test数据集
X_train, X_test, y_train, y_test = train_test_split(X, y, random_state=0)

cmap = cm.get_cmap('gnuplot')
scatter = pd.scatter_matrix(X_train, c= y_train, marker = 'o', s=40, hist_kwds={'bins':15}, figsize=(9,9), cmap=cmap)

# 画3D点图
from mpl_toolkits.mplot3d import Axes3D

fig = plt.figure()
ax = fig.add_subplot(111, projection = '3d')
ax.scatter(X_train['width'], X_train['height'], X_train['color_score'], c = y_train, marker = 'o', s=100)
ax.set_xlabel('width')
ax.set_ylabel('height')
ax.set_zlabel('color_score')
plt.show()

# 把mass,width,height作为决定因素
X = fruits[['mass', 'width', 'height']]
y = fruits['fruit_label']

# 默认 75% / 25% train-test split
X_train, X_test, y_train, y_test = train_test_split(X, y, random_state=0)

#创建knn分类器
1 from sklearn.neighbors import KNeighborsClassifier
#n_neighbors分类数量
knn = KNeighborsClassifier(n_neighbors = 5)

#数据拟合
knn.fit(X_train, y_train)

KNeighborsClassifier(algorithm='auto', leaf_size=30, metric='minkowski',
           metric_params=None, n_jobs=1, n_neighbors=5, p=2,
           weights='uniform')


使用test集来计算模型的准确性

knn.score(X_test, y_test)

0.53333333333333333

使用训练好的knn模型来预测新数据

# first example: a small fruit with mass 20g, width 4.3 cm, height 5.5 cm
fruit_prediction = knn.predict([[20, 4.3, 5.5]])
lookup_fruit_name[fruit_prediction[0]]

'mandarin'

# second example: a larger, elongated fruit with mass 100g, width 6.3 cm, height 8.5 cm
fruit_prediction = knn.predict([[100, 6.3, 8.5]])
lookup_fruit_name[fruit_prediction[0]]

'lemon'

画knn的决策边界

from adspy_shared_utilities import plot_fruit_knn

plot_fruit_knn(X_train, y_train, 5, 'uniform')   # we choose 5 nearest neighbors

knn中聚类个数k对最后准确率的影响

k_range = range(1,20)
scores = []

for k in k_range:
    knn = KNeighborsClassifier(n_neighbors = k)
    knn.fit(X_train, y_train)
    scores.append(knn.score(X_test, y_test))

plt.figure()
plt.xlabel('k')
plt.ylabel('accuracy')
plt.scatter(k_range, scores)
plt.xticks([0,5,10,15,20]);

knn对数据集train/test划分比例的敏感性

t = [0.8, 0.7, 0.6, 0.5, 0.4, 0.3, 0.2]

knn = KNeighborsClassifier(n_neighbors = 5)

plt.figure()

for s in t:

    scores = []
    for i in range(1,1000):
        X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = 1-s)
        knn.fit(X_train, y_train)
        scores.append(knn.score(X_test, y_test))
    plt.plot(s, np.mean(scores), 'bo')

plt.xlabel('Training set proportion (%)')
plt.ylabel('accuracy');