柚子快報邀請碼778899分享：聚類 機(jī)器學(xué)習(xí) 深度學(xué)習(xí) K

http://yzkb.51969.com/

主要用于保存記錄，來源B站視頻跟著大佬做的。

實(shí)現(xiàn)效果：

1.原始數(shù)據(jù)分類? 2.原始數(shù)據(jù)未分類

1.原始數(shù)據(jù)分類? ?2.數(shù)據(jù)聚類后（‘x’是每個聚類的中心）

demo:

### demo

import numpy as np

import pandas as pd

import matplotlib.pyplot as plt

from k_means import KMeans

# 導(dǎo)入數(shù)據(jù)

data = pd.read_csv("./iris.csv")

iris_types = ['setosa','versicolor','virginica'] # 數(shù)據(jù)類別

x_axis = 'Petal_Length' #x軸

y_axis = 'Petal_Width' #Y軸

# 圖的大小

plt.figure(figsize=(12,5))

plt.subplot(1,2,1)

for iris_type in iris_types: # 遍歷每種花

plt.scatter(data[x_axis][data['Species']==iris_type],data[y_axis][data['Species']==iris_type],label=iris_type)

plt.title('label know')

plt.legend()

plt.subplot(122)

plt.scatter(data[x_axis][:],data[y_axis][:])

plt.title('label unknow')

plt.show()

num_examples = data.shape[0] #樣本個數(shù)

x_train = data[[x_axis,y_axis]].values.reshape(num_examples, 2)#將data的x,y軸數(shù)據(jù)轉(zhuǎn)換成ndarry格式，并reshape，賦給x_train

#指定好訓(xùn)練所需的參數(shù)

num_clusters = 3

max_iterition = 50

KMeans = KMeans(x_train,num_clusters)

centroids, closest_centroids_ids = KMeans.train(max_iterition)

#畫圖對比

plt.figure(figsize=(12,5)) # 圖的大小

plt.subplot(1,2,1)

for iris_type in iris_types: # 遍歷每種花

plt.scatter(data[x_axis][data ['Species']==iris_type],data[y_axis][data['Species']==iris_type],label=iris_type)

plt.title('label know')

plt.legend()

plt.subplot(1,2,2)

for centroids_id, centroid in enumerate(centroids):

current_examples_index = (closest_centroids_ids == centroids_id).flatten()

plt.scatter(data[x_axis][current_examples_index], data[y_axis][current_examples_index], label=centroids_id)

for centroids_id, centroid in enumerate(centroids):

plt.scatter(centroid[0],centroid[1],c='black',marker='x')

plt.title('label KMeans')

plt.legend()

plt.show()

K_means.py:

### K_means

import numpy as np

# closest_centroids_ids里面存的是什么？ ——>存的是每個數(shù)據(jù)距離中心點(diǎn)最小距離的簇的索引(1,2,3,...,K)

class KMeans:

def __init__(self, data, num_clustres): # 數(shù)據(jù)、K值

self.data = data

self.num_clustres = num_clustres

def train(self, max_iterations): # 最大迭代次數(shù)

# 1在當(dāng)前數(shù)據(jù)中，初始化隨機(jī)選擇K個中心點(diǎn)

centroids = KMeans.centroids_init(self.data, self.num_clustres)

# 2開始訓(xùn)練

# 計算每個點(diǎn)到K個中心點(diǎn)的距離

num_examples = self.data.shape[0] # 數(shù)據(jù)個數(shù)

closest_centroids_ids = np.empty((num_examples, 1)) # 創(chuàng)建空數(shù)組，存放尋找的最近的中心點(diǎn)

for _ in range(max_iterations): # 迭代max_iterations次

# 3得到當(dāng)前每個樣本點(diǎn)到K個中心點(diǎn)的距離，找到最近的距離id

closest_centroids_ids = KMeans.centroids_find_closest(self.data, centroids) # 數(shù)據(jù)、中心點(diǎn)

# 4進(jìn)行中心點(diǎn)位置更新：closest_centroids_ids距離誰最近就屬于哪個堆

centroids = KMeans.centroids_computer(self.data, closest_centroids_ids, self.num_clustres) #數(shù)據(jù)、當(dāng)前數(shù)據(jù)距離哪個點(diǎn)最近

return centroids, closest_centroids_ids

@staticmethod

# num_clustres()方法：初始化隨機(jī)的尋找當(dāng)前data的num_clustres個中心點(diǎn)

def centroids_init(data, num_clustres):

num_examples = data.shape[0] # 數(shù)據(jù)個數(shù)

random_ids = np.random.permutation(num_examples) # permutation：將數(shù)據(jù)隨機(jī)排序

# random_ids[:num_clustres]：選擇從0到num_clustres個數(shù)據(jù)，前num_clustres個；后面的：表示所有的數(shù)據(jù)特征

data = np.array(data)

centroids = data[random_ids[:num_clustres], :] # 選取后作為中心點(diǎn)

return centroids

@staticmethod

# 計算距離中心的的最近的距離，存儲在closest_centroids_ids中，返回

def centroids_find_closest(data, centroids):

# 計算方法：eg.歐氏距離

num_examples = data.shape[0] # 數(shù)據(jù)個數(shù)

num_centroids = centroids.shape[0] # 簇的個數(shù)：K

closest_centroids_ids = np.zeros((num_examples, 1)) # 用于存儲每個樣本點(diǎn)距離簇最近的那一個

for examples_index in range(num_examples):

distance = np.zeros((num_centroids, 1)) # 存儲每個樣本點(diǎn)對K個簇的distance值

for centroids_index in range(num_centroids):

# 計算距離：數(shù)據(jù)點(diǎn)的坐標(biāo)值-中心點(diǎn)的坐標(biāo)值

data = np.array(data)

distance_diff = data[examples_index, :4] - centroids[centroids_index, :4]

distance[centroids_index] = np.sum(distance_diff ** 2)

# 對當(dāng)前的樣本找到距離最近的那個簇

closest_centroids_ids[examples_index] = np.argmin(distance) # np.argmin()返回最小值索引

return closest_centroids_ids

@staticmethod

# 更新closest_centroids_ids里面距離誰最近，就放在哪個堆（簇）

def centroids_computer(data, closest_centroids_ids, num_clustres):

num_features = data.shape[1] # 輸入的數(shù)據(jù)的特征個數(shù)

# num_features: 用于各個特征分別算均值

centroids = np.zeros((num_clustres,num_features)) # 簇的個數(shù)、特征個數(shù)

for centroids_id in range(num_clustres):

###如何運(yùn)行的？

closest_ids = closest_centroids_ids == centroids_id

centroids[centroids_id] = np.mean(data[closest_ids.flatten(),:], axis=0)

return centroids

柚子快報邀請碼778899分享：聚類 機(jī)器學(xué)習(xí) 深度學(xué)習(xí) K

http://yzkb.51969.com/

參考閱讀