K-means聚类

目录

K-means聚类

1 简介

2 Python实战

---

1 简介

原理：通过计算不同样本间的距离来判断他们的相近关系的，相近的就会放到同一个类别中去。

适用数据：数值数据

优点：思想简单，容易实现，可解释度比较强

缺点：对噪音和异常点比较的敏感。k-means是在做凸优化，因此处理不了非凸的分布,对于条形或不规则形状的数据，效果较差。如果两个类别距离比较近，k-means的效果也不会太好。初始中心点的选择以及k值的选择对结果影响较大，可能每次聚类结果都不一样。结果可能只是局部最优而不是全局最优

2 Python实战

数据集:天池学习赛数据集——汽车产品聚类分析（有兴趣的同学可以直接参赛学习一下）

任务：对该汽车数据进行聚类分析，并找到vokswagen汽车的相应竞品。

数据介绍：car_price.csv，数据包括了205款车的26个字段

1	Car_ID	Unique id of each observation (Interger)
2	Symboling	Its assigned insurance risk rating, A value of +3 indicates that the auto is risky, -3 that it is probably pretty safe.(Categorical)
3	carCompany	Name of car company (Categorical)
4	fueltype	Car fuel type i.e gas or diesel (Categorical)
5	aspiration	Aspiration used in a car (Categorical)
6	doornumber	Number of doors in a car (Categorical)
7	carbody	body of car (Categorical)
8	drivewheel	type of drive wheel (Categorical)
9	enginelocation	Location of car engine (Categorical)
10	wheelbase	Weelbase of car (Numeric)
11	carlength	Length of car (Numeric)
12	carwidth	Width of car (Numeric)
13	carheight	height of car (Numeric)
14	curbweight	The weight of a car without occupants or baggage. (Numeric)
15	enginetype	Type of engine. (Categorical)
16	cylindernumber	cylinder placed in the car (Categorical)
17	enginesize	Size of car (Numeric)
18	fuelsystem	Fuel system of car (Categorical)
19	boreratio	Boreratio of car (Numeric)
20	stroke	Stroke or volume inside the engine (Numeric)
21	compressionratio	compression ratio of car (Numeric)
22	horsepower	Horsepower (Numeric)
23	peakrpm	car peak rpm (Numeric)
24	citympg	Mileage in city (Numeric)
25	highwaympg	Mileage on highway (Numeric)
26	price(Dependent variable)	Price of car (Numeric)

数据集以及源码下载：链接：https://pan.baidu.com/s/1yGpMJIu_vS31GjM8mc_8UA 提取码：5n6x 

代码以及结果展示：

import pandas as pd

# 数据加载
data = pd.read_csv('./car_price.csv')
data

# 使用KMeans进行聚类,导入库
from sklearn.cluster import KMeans
#预处理
from sklearn import preprocessing
from sklearn.preprocessing import LabelEncoder
import pandas as pd
#矩阵运算
import numpy as np

# 建立训练集
train_x = data[["car_ID","symboling","CarName", "fueltype", "aspiration", "doornumber",
                    "carbody", "drivewheel", "enginelocation", "wheelbase", "carlength","carwidth",
                    "carheight", "curbweight", "enginetype", "cylindernumber", "enginesize","fuelsystem",
                    "boreratio", "stroke", "compressionratio", "horsepower", "peakrpm","citympg",
                    "highwaympg", "price"]]
train_x

#将非数值字段转化为数值
le = LabelEncoder()
columns = ['CarName','fueltype','aspiration','doornumber','carbody','drivewheel','enginelocation','enginetype','cylindernumber','fuelsystem']
for column in columns:
    #LabelEncoder().transform将标签转换为归一化的编码。fit_transform安装标签编码器并返回编码的标签。
    train_x[column] = le.fit_transform(train_x[column])
train_x

# 规范化到 [0,1] 空间
min_max_scaler=preprocessing.MinMaxScaler()
#MinMaxScaler()将每个要素缩放到给定范围，拟合数据，然后对其进行转换。
train_x=min_max_scaler.fit_transform(train_x)
pd.DataFrame(train_x).to_csv('temp.csv', index=False)

#选择聚类组数
import matplotlib.pyplot as plt
sse = []
for k in range(1, 11):
	kmeans = KMeans(n_clusters=k)
	kmeans.fit(train_x)
	# 计算inertia簇内误差平方和
	sse.append(kmeans.inertia_)
x = range(1, 11)
#将图像嵌入在结果中
%matplotlib inline
plt.xlabel('K')
plt.ylabel('SSE')
plt.plot(x, sse, 'o-')

# 使用KMeans聚类,分成4类
kmeans = KMeans(n_clusters=5)
kmeans.fit(train_x)  # 也可以直接fit+predict
#predict计算聚类中心并预测每个样本的聚类索引。
predict_y = kmeans.predict(train_x)
predict_y

# 合并聚类结果，插入到原数据中,axis： 需要合并链接的轴，0是行，1是列 
result = pd.concat((data,pd.DataFrame(predict_y)),axis=1)
# 将结果列重命名为'聚类结果'
result.rename({0:u'聚类结果'},axis=1,inplace=True)
result

# 找出vokswagen汽车的聚类结果
label = result[result.CarName.str.contains('vokswagen')]['聚类结果']
label
#以下此段仅为练习，可忽略。多个关键词搜索，join() 方法用于将序列中的元素（必须是str）以指定的字符连接生成一个新的字符串。
List = ['vokswagen','volkswagen']
mask = result.CarName.str.contains('|'.join(List))
selected_data = result[mask]
selected_data

# Vokswagen 轿车 竞品按价格排序
#lambda 是为了减少单行函数的定义而存在的,lambda作为一个表达式，定义了一个匿名函数，上例的代码x为入口参数，x['聚类结果']==3and 'sedan' in x['carbody']为函数体，
#将函数应用到由各列或行形成的一维数组上。DataFrame的apply方法可以实现此功能。默认情况下会以列为单位,axis = 1以行为单位
#[[]]选择多列时用双括号
result[result.apply(lambda x : x['聚类结果'] ==3 and 'sedan' in x['carbody'],axis = 1)][['CarName',"wheelbase", "price",'horsepower','carbody','fueltype','聚类结果']].sort_values('price',ascending = False)

# Vokswagen wagon 竞品按价格排序
result[result.apply(lambda x : x['聚类结果'] ==3 and 'wagon' in x['carbody'],axis =1)][['CarName',"wheelbase", "price",'horsepower','carbody','fueltype','聚类结果']].sort_values('price',ascending = False)

#显示竞品车总体聚类结果
benchmark = result[result['聚类结果']==3].CarName
print("竞品车如下所示")
print(benchmark)

注明：以上代码仅为学习，代码参考比赛论坛的某帖子，实名感谢，另外还有很多不完善的地方，大家可以提出，一起探讨。