一、实验目的：

掌握数据集成、变换、归约等预处理技术方法；
掌握复杂数据对象的相似性度量方法

二、实验内容：

1.基础数据预处理技术
集成Wine1和Wine2两个数据集，形成新数据集Wine
将Wine中的Alcohol 和Proline两个属性标准化处理，映射到[0,1]区间
根据Alcohol（%）和 Weight（g）两个属性计算每份样本中的酒精质量
2.复杂数据对象的相似性度量
创建数据集，数据集中包含10个数据对象，每个数据对象由一个二值型属性和两个五维向量组成，如下所示

属性A的权重为0.3，属性B的权重为0.2，属性C的权重为0.5
属性A的不同属性值间由简单匹配系数度量相似性
属性B的不同属性值间由欧式距离度量距离
属性C的不同属性值间由余弦距离度量相似性
计算不同数据对象间相似性，以矩阵形式输出计算结果
（相似性与距离转换函数：s=1/(1+d)）

三、实验设备及软件环境：

Windows10
Python3.8
PyCharm 2020.3.5 (Professional Edition)

四、实验过程及结果：

Shiyan1.1.py

import xlrd
import xlwt


def read():
    data1 = xlrd.open_workbook(r'C:\Users\zlh\Desktop\数据分析方法实验\实验一、二\实验一数据\wine1.xlsx')
    sheet1 = data1.sheet_names()
    table1 = data1.sheet_by_name(sheet1[0])

    data2 = xlrd.open_workbook(r'C:\Users\zlh\Desktop\数据分析方法实验\实验一、二\实验一数据\wine2.xlsx')
    sheet2 = data2.sheet_names()
    table2 = data2.sheet_by_name(sheet2[0])

    Alcohol = table1.col_values(1)
    del Alcohol[0]
    # print(Alcohol)
    Weight = table1.col_values(5)
    del Weight[0]
    Proline = table2.col_values(8)
    del Proline[0]
    return table1, table2, Alcohol, Weight, Proline


def write(table1, table2, Alcohol, Weight, Proline, p, a):
    workbook = xlwt.Workbook(encoding='utf-8')
    worksheet = workbook.add_sheet('Sheet1')

    for i in range(table1.ncols):
        for j in range(table1.nrows):
            worksheet.write(j, i, label=table1.cell(j, i).value)
    for i in range(1, table2.ncols):
        for j in range(table2.nrows):
            worksheet.write(j, i + table1.ncols - 1, label=table2.cell(j, i).value)

    Quality = ["Quality"]
    for i in range(len(Alcohol)):
        n = Alcohol[i] / Weight[i]
        Quality.append(n)

    for j in range(table2.nrows):
        worksheet.write(j, 14, label=p[j])
        worksheet.write(j, 15, label=a[j])
        worksheet.write(j, 16, label=Quality[j])

    workbook.save('Wine.xls')


def Standardization(Alcohol, Proline):
    Proline_min = min(Proline)
    Proline_max = max(Proline)
    Alcohol_min = min(Alcohol)
    Alcohol_max = max(Alcohol)
    # print(Proline_min)
    # print(Proline_max)
    # print(Alcohol_min)
    # print(Alcohol_max)
    Alcohol_Standardization = ['Alcohol_Standardization']
    Proline_Standardization = ['Proline_Standardization']
    for i in range(len(Alcohol)):
        num = (Alcohol[i] - Alcohol_min) / (Alcohol_max - Alcohol_min)
        Alcohol_Standardization.append(num)
        # print(num)
    for i in range(len(Proline)):
        num = (Proline[i] - Proline_min) / (Proline_max - Proline_min)
        Proline_Standardization.append(num)
        # print(num)
    return Proline_Standardization, Alcohol_Standardization


if __name__ == "__main__":
    table1, table2, Alcohol, Weight, Proline = read()
    p, a = Standardization(Alcohol, Proline)
    write(table1, table2, Alcohol, Weight, Proline, p, a)

Shiyan1.2.py

import math
from random import randrange, random


A=[[0 for i in range(6)]for i in range(10)]
B = [[0 for i in range(6)]for i in range(10)]
C = [[0 for i in range(6)]for i in range(10)]
for i in range(0, 10):
    for j in range(0, 6):
        if random() < 0.5:
            A[i][j] = 0
        else:
            A[i][j] = 1
        B[i][j] = randrange(0, 10)
        C[i][j] = randrange(0, 10)


def Smc(f00, f01, f10, f11):
    smc = [[0 for i in range(10)]for i in range(10)]
    smc1 = [[0 for i in range(10)]for i in range(10)]
    for j in range(0, 10):
        for i in range(0, 10):
            smc[j][i] = (f11[j][i] + f00[j][i])
            smc1[j][i] = smc[j][i] / (f01[j][i] + f10[j][i] + f11[j][i] + f00[j][i])
    return smc1


def distB():
    sumb =[[0 for i in range(10)]for i in range(10)]
    sb = [[1 for i in range(10)]for i in range(10)]
    distb = [[0 for i in range(10)]for i in range(10)]
    for n in range(0, 10):
        for xb in range(0, 10):
            for yb in range(0, 6):
                sumb[n][xb] = sumb[n][xb] + (B[n][yb] - B[xb][yb]) * (B[n][yb] - B[xb][yb])
            # print(sumb)
            distb[n][xb] = math.sqrt(sumb[n][xb])
            # print(distb)
            sb[n][xb] = sb[n][xb] / (1 + distb[n][xb])
    return sb


def smcC():
    sumc1 = [[0 for i in range(10)]for i in range(10)]
    sumc2 = [[0 for i in range(10)]for i in range(10)]
    sumc3 = [[0 for i in range(10)]for i in range(10)]
    sumc = [[0 for i in range(10)]for i in range(10)]
    sum = [[0 for i in range(10)]for i in range(10)]

    for n in range(0, 10):
        for xc in range(0, 10):
            for yc in range(0, 6):
                sumc1[n][xc] = sumc1[n][xc] + C[n][yc] * C[xc][yc]
            for yc in range(0, 6):
                sumc2[n][xc] = sumc2[n][xc] + C[n][yc] * C[n][yc]
            for yc in range(0, 6):
                sumc3[n][xc] = sumc3[n][xc] + C[xc][yc] * C[xc][yc]
            sumc[n][xc] = math.sqrt(sumc2[n][xc]) * math.sqrt(sumc3[n][xc])
            sum[n][xc] = sumc1[n][xc] / sumc[n][xc]
    return sum


def Sim(a, b, c, wa, wb, wc):
    sim = [[0 for i in range(10)]for i in range(10)]

    for n in range(0, 10):
        for i in range(0, 10):
            sim[n][i] = wa * a[n][i] + wb * b[n][i] + wc * c[n][i]
    return sim


def f():
    f00 = [[0 for i in range(10)]for i in range(10)]
    f01 = [[0 for i in range(10)]for i in range(10)]
    f10 = [[0 for i in range(10)]for i in range(10)]
    f11 = [[0 for i in range(10)]for i in range(10)]

    for n in range(0, 10):
        for j in range(0, 10):
            for i in range(0, 6):
                if (A[n][i] == A[j][i]) & A[n][i] == 0:
                    f00[n][j] = f00[n][j] + 1

            for i in range(0, 6):
                if (A[n][i] != A[j][i]) & A[n][i] == 0 & A[j][i] == 1:
                    f01[n][j] = f01[n][j] + 1
            # print(f01)
            for i in range(0, 6):
                if (A[n][i] != A[j][i]) & A[n][i] == 1:
                    f10[n][j] = f10[n][j] + 1
            # print(f10)
            for i in range(0, 6):
                if (A[n][i] == A[j][i]) & A[n][i] == 1:
                    f11[n][j] = f11[n][j] + 1
            # print(f11)
    # print(f00)
    # print(f01)
    # print(f10)
    # print(f11)
    return f00, f01, f10, f11

def out(e):
    for i in range(0,10):
        for j in range(0,10):
            print('%-10f' % (e[i][j]),end=" ")
        print()

if __name__ == "__main__":
    f00, f01, f10, f11 = f()
    sA = Smc(f00, f01, f10, f11)
    print("简单匹配系数度量相似性")
    out(sA)
    sB = distB()
    print("欧式距离度量距离")
    out(sB)
    sC = smcC()
    print("余弦距离度量相似性")
    out(sC)
    sim = Sim(sA, sB, sC, 0.3, 0.2, 0.5)
    print("相似性")
    out(sim)

运行结果：

实验1.1

实验1.2

五、总结：

通过本次实验，掌握数据了集成、变换、归约等预处理技术方法，掌握了复杂数据对象的相似性度量方法

实验成绩： 指导教师： 年 月 日