python 朴素贝叶斯模拟联邦学习无框架

数据集：https://archive.ics.uci.edu/ml/datasets/Adult

数据集本身是做二分类的。

里面自己设计了DataLoader、Tester等类，简单易懂。

导入的头文件都没卵用，全是手写的。

tester用了从30000条到32000条数据，当然你自己可以随便改。

Party类实例化之后是多方中的一方。

import torch
import torch.nn.modules
import torch.nn
import numpy as np
from torch.autograd import Variable #torch的基本变量
import torch.nn.functional as F #里面有很多torch的函数
import matplotlib.pyplot as plt

NUM_OF_TEST_SET=2000
NUM_OF_PARTY=5
PARTY_DATA=6000
NUM_OF_ALL_DATA=32561

class Party:
    def __init__(self):
        self.Data=[]
        self.NUMOFFEATURE=0
        self.NUMOFDATA=0
    def Update(self):
        self.NUMOFDATA=len(self.Data)
        self.NUMOFFEATURE=len(self.Data[0])
    def BayesPredict(self,feature):
        #use=self.ConvertFeature(feature)
        #if len(use)==0: return "Feature Convert Failed. Unable to Predict"
        #print('one')

        half0=[item[:-1] for item in self.Data if item[self.NUMOFFEATURE-1]==1]#大于50K的
        cnt=[0]*len(feature)
        for item in half0:
            for index,data in enumerate(item):
                if data==feature[index]:
                    cnt[index]+=1
        cnt=cnt[:-1]
        P1 = 1
        PA=len(half0)#0类分类在总分类中出现的次数
        if PA!=0:
            #特征事件相乘
            for count in cnt:
                P1 =  P1*(count/PA)
            #分类事件相乘
            P1 = P1 * (PA/self.NUMOFDATA)


        half1 = [item[:-1] for item in self.Data if item[self.NUMOFFEATURE - 1] == 0]  # 小于50K的
        cnt = [0] * len(feature)
        for item in half1:
            for index, data in enumerate(item):
                if data == feature[index]:
                    cnt[index] += 1
        cnt = cnt[:-1]
        P2 = 1
        PA = len(half1)  # 0类分类在总分类中出现的次数
        if PA!=0:
            # 特征事件相乘
            for count in cnt:
                P2 = P2 * (count / PA)
            # 分类事件相乘
            P2 = P2 * (PA / self.NUMOFDATA)
        if P1-P2<1e-11: return -1
        elif P1>P2: return 1#大于50k
        elif P1<P2: return 0#小于50k


    def ConvertFeature(self,feature):
        res=[]
        for item in feature:
            print(item)
            if item in self.GetClass:
                res.append(self.GetClass[item])
            else:return []
        return res

class DataLoader():
    def __init__(self):
        self.RawData = []
        self.Data = []
        self.GetClass = {}

        self.NUMOFFEATURE = 0
        self.NUMOFDATA = 0
    def LoadData(self,PATH):
        with open(PATH,'r') as f:
            for line in f:
                self.RawData.append([thing.strip() for thing in line.split(',')])
        self.RawData=self.RawData[:-1]
        self.NUMOFDATA=len(self.RawData)
    def ShowRawData(self):
        for item in self.RawData:
            print(item)
    def WashData(self):
        attribute=len(self.RawData[0])-1
        #print("attribute: {}".format(attribute))

        index=0
        while index<attribute+1:
            #print('index: {}'.format(index))
            cnt=0
            temp=[i[index] for i in self.RawData]

            for item in temp:
                if item not in self.GetClass:
                    self.GetClass[item]=cnt
                    cnt+=1
            index+=1
        self.NUMOFFEATURE=index

        #print("feature number is {}.".format(self.NUMOFFEATURE))

        for line in self.RawData:
            self.Data.append([self.GetClass[i] for i in line])
    def ShowData(self):
        for item in self.Data:
            print(item)
    def AssignData(self,Party,start,end):
        Party.Data=self.Data[start:end]

class Tester:
    def __init__(self):
        self.Data = []

        self.NUMOFFEATURE = 0
        self.NUMOFDATA = 0
    def Update(self):
        self.NUMOFDATA=len(self.Data)
        self.NUMOFFEATURE=len(self.Data[0])
    def Test(self,A):
        Error=0
        #print(A.Data[:10])
        for item in self.Data:
            res=A.BayesPredict(item)
            if res==-1:continue
            if res!=item[-1]:
                Error+=1
        return Error


if __name__=="__main__":
    FilePath=r'C:\Users\Lenovo\Desktop\adult.data'



    #simulation of federated learning
    loader=DataLoader()
    loader.LoadData(FilePath)
    loader.WashData()
    #print(loader.GetClass['>50K'])
    #print(loader.GetClass['<=50K'])

    parties=[]
    for i in range(6):
        temp=Party()
        parties.append(temp)



    last=0
    for i in range(len(parties)):
        loader.AssignData(parties[i],last,last+6000)
        #print(parties[i].Data[:10])
        parties[i].Update()
        last+=6000

    tester = Tester()
    loader.AssignData(tester,31000,32000)
    '''
    y=[]
    last=0
    for i in range(6):
        loader.AssignData(parties[0],last,last+6000)
        res=tester.Test(parties[0])
        y.append(res)
    '''
    y=[]

    last = 6000
    for i in range(6):
        loader.AssignData(parties[0], 0, last)
        res = tester.Test(parties[0])
        y.append(res)

    x = [1, 2, 3, 4, 5, 6]
    plt.bar(x, y)
    plt.show()