机器学习 贝叶斯估计学习笔记

数据准备

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
%matplotlib inline

from sklearn.datasets import load_iris
from sklearn.model_selection import train_test_split

from collections import Counter
import math

# data
def create_data():
    iris = load_iris()
    df = pd.DataFrame(iris.data, columns=iris.feature_names)
    df['label'] = iris.target
    df.columns = ['sepal length', 'sepal width', 'petal length', 'petal width', 'label']
    data = np.array(df.iloc[:100, :])
    # print(data)
    return data[:,:-1], data[:,-1]

X, y = create_data()
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3)

在这里插入代码片

核心代码

class NaiveBayes:
    def __init__(self):
        self.model = None

    # 数学期望
    @staticmethod
    def mean(X):
        #X是传入的特征向量(此处为二维)
        return sum(X) / float(len(X))

    # 标准差（方差）
    def stdev(self, X):
        avg = self.mean(X)
        return math.sqrt(sum([pow(x - avg, 2) for x in X]) / float(len(X)))

    # 概率密度函数
    def gaussian_probability(self, x, mean, stdev):
        exponent = math.exp(-(math.pow(x - mean, 2) /
                              (2 * math.pow(stdev, 2))))
        return (1 / (math.sqrt(2 * math.pi) * stdev)) * exponent

    # 处理X_train
    def summarize(self, train_data):
        summaries = [(self.mean(i), self.stdev(i)) for i in zip(*train_data)]
        #此处将X_train(train_data)拆解为100个列表
        #将每一个数据,例如[3,4,...]，求出它的期望和方差，得到一个新的列表
        return summaries

    # 分类别求出数学期望和标准差
    #fit():训练模型，得到参数
    def fit(self, X, y):
        labels = list(set(y))
        #得到的labels=[0,1](前一百个数据的标签)
        #iris[字典]
		#df[DataFrame]
		#data[ndarray]
		#set(y)[set(集合)]
		
        data = {label: [] for label in labels}
        #字典推导式 label是key,[]是value 
        #data:{0.0: [], 1.0: []}
        
        for f, label in zip(X, y):
            data[label].append(f)
        #这个for循环：data:{0.0: [[1,2,...],[3,4,...],...], 1.0: [[],[],...]}
        
        #model是字典，label做键，[(均值of data1,方差of data1),(均值of data2,方差of data2)]做值
        self.model = {
            label: self.summarize(value)
            for label, value in data.items()
            #遍历data
            #data.items()：
            #dict_items([(1, [5,4,6,7,...]), (0, [6,9,1,2,...])])
            # summaries:{0.0: [(5.0, 0.37),(3.42, 0.40)], 1.0: [(5.8, 0.449),(2.7, 0.27)]}
            #model{0:[(5.0, 0.37),(3.42, 0.40)],1.0:[(5.8, 0.449),(2.7, 0.27)]}
                    }
            
                    
        return 'gaussianNB train done!'

    # 计算概率
    def calculate_probabilities(self, input_data):
        # summaries:{0.0: [(5.0, 0.37),(3.42, 0.40)], 1.0: [(5.8, 0.449),(2.7, 0.27)]}
        # input_data:[1.1, 2.2,3,4]
        # input_data是一个多维向量,input_data=x_test
        # model.model.items()：
        # (0.0, [(4.96875, 0.3786798614925278), (3.328125000000001, 0.37519526166384365), (1.4562499999999996, 0.16189792308735776), (0.2468750000000001, 0.09995115994824672)]),
        # 1.0, [(5.91578947368421, 0.4676516448616729), (2.78421052631579, 0.30480909969621606), (4.292105263157894, 0.4521187425231565), (1.3447368421052632, 0.19289671141374656)])]
 )
        #iris数据集有四个特征向量!
        probabilities = {}
        
        #这个嵌套for循环对每一个特征向量进行操作
        for label, value in self.model.items():
            probabilities[label] = 1
            #probabilities = {1:1,0:1}
            
            #这个for循环对每一个特征向量，计算它的概率密度
            for i in range(len(value)):
                mean, stdev = value[i]#来自model的均值和方差
                probabilities[label] *= self.gaussian_probability(
                    input_data[i], mean, stdev)#x_test数据
                   
        return probabilities
		#probabilities={1:[0.9999,0.8],0:[0.2,0.6]}
    # 类别
    def predict(self, X_test):
        # {0.0: 2.9680340789325763e-27, 1.0: 3.5749783019849535e-26}
        label = sorted(
            self.calculate_probabilities(X_test).items(),
            key=lambda x: x[-1])[-1][0]
            
        return label

    def score(self, X_test, y_test):
        right = 0
        for X, y in zip(X_test, y_test):
            label = self.predict(X)
            if label == y:
                right += 1

        return right / float(len(X_test))

for label, value in self.model.items():
            probabilities[label] = 1
         
            for i in range(len(value)):
                mean, stdev = value[i]
                probabilities[label] *= self.gaussian_probability(
                    input_data[i], mean, stdev)

重点！！：
分label=0遍历一次四元素二元组，和label=1遍历一次四元素二元组。
len(value)=4

>>model.model.items()
>>dict_items(
 (0.0, [(4.96875, 0.3786798614925278), (3.328125000000001, 0.37519526166384365), (1.4562499999999996, 0.16189792308735776), (0.2468750000000001, 0.09995115994824672)]),
 (1.0, [(5.91578947368421, 0.4676516448616729), (2.78421052631579, 0.30480909969621606), (4.292105263157894, 0.4521187425231565), (1.3447368421052632, 0.19289671141374656)])]
 )
 >>model.calculate_probabilities([1.11111, 2.2,3,4])
 >>{0.0: 0.0, 1.0: 4.691960853801238e-67}
 >>import pandas as pd
 >>pd.DataFrame(model.model.values())

zip(*a)

• zip(*a)与zip(a)相反，理解为解压

>>> a = list(zip([1,2,3],[4,5,6],[4,5,6,7,8]))
>>> a
[(1, 4, 4), (2, 5, 5), (3, 6, 6)]
>>> c,d,e = zip(*a)
>>> c
(1, 2, 3)
>>> d
(4, 5, 6)
>>> e
(4, 5, 6)

set()

• Python 中 set 基本特点：
  (1) 无序性
  (2) 确定性
  (3) 不重复性
• Python 中 set() 实质：
  内部进行 可迭代性的 for 循环
• 创建set:

>>> set([1,2,3])
{1, 2, 3}
>>> set('123')
{'1', '2', '3'}
>>> set()
set() #创建一个空set

>>>set("Hello")
{'H', 'e', 'l', 'o'}

• 增加一个元素：
  add()用于增加一个元素值，

  update()：是把要传入的元素拆分，做为个体传入到集合中

  update([])，用于增加多个元素值，参数为list，注意如果用add增加多个值，会报参数类型错误。

>>> a = set()
>>> a.add('python')
>>> a
{'python'}
>>> a.add('love')
>>> a
{'love', 'python'}
>>> a.add('i')
>>> a
{'love', 'i', 'python'}

>>> a = set('love')
>>> a
{'l', 'v', 'o', 'e'}
>>> a.update('python')
>>> a
{'h', 'y', 'e', 'n', 'o', 'p', 't', 'l', 'v'}
>>> a.update(['you','me'])
 {'e', 'h', 'l', 'me', 'n', 'o', 'p', 't', 'v', 'y', 'you'}

字典的.append()

a = (0,0)
b = ('a','b')
d = dict(zip(b,a))
l = []
for i in [1, 2, 3]:
    d['a'] = i
    d['b'] = i*2
    l.append(d)
print (l)

[{'a': 3, 'b': 6}, {'a': 3, 'b': 6}, {'a': 3, 'b': 6}]

items()

字典里对应的一对键和值以元组的形式（键, 值），存储为所生成序列里的单个元素

favorite_languages = {
    'jen':[1,2,3],
	'sarah':'c',
	'edward':'rby',
	'phil':'python',
	}
test_items = favorite_languages.items()
print(test_items)
print('\n测试items()方法的返回值是不是list类\n')
print(isinstance(test_items,list))
print('\n测试items()方法的返回值是哪种类\n')
print(type(test_items))
print('\n测试dict_items类的每一个元素是哪种类\n')
for i in test_items:
    print(i)
    print(type(i))

dict_items([('jen', [1, 2, 3]), ('sarah', 'c'), ('edward', 'rby'), ('phil', 'python')])

测试items()方法的返回值是不是list类

False

测试items()方法的返回值是哪种类

<class 'dict_items'>

测试dict_items类的每一个元素是哪种类

('jen', [1, 2, 3])
<class 'tuple'>
('sarah', 'c')
<class 'tuple'>
('edward', 'rby')
<class 'tuple'>
('phil', 'python')
<class 'tuple'>

dict 妙用

probabilities = {}
probabilities[0] = 1
probabilities[1] = 2
probabilities

{0: 1, 1: 2}

高斯模型

• 高斯分布：正态分布，一种连续型变量的概率分布。
• 高斯概率分布是由均值μ和标准差σ唯一确定的。
  在这里插入图片描述
  
• 在贝叶斯分类中，高斯模型就是用来处理连续型特征变量的，当使用此模型时，我们会假定特征属于高斯分布，然后基于训练样本计算特征均值和标准差，这样就可以得到此特征下每一个属性值的先验概率。