基于Sklearn机器学习实战---基于Sklearn模块的链路预测

Sklearn简介

自2007年发布以来，scikit-learn已经成为Python重要的机器学习库了。scikit-learn简称sklearn，支持包括分类、回归、降维和聚类四大机器学习算法。还包含了特征提取、数据处理和模型评估三大模块。

  sklearn是Scipy的扩展，创建在NumPy和matplotlib库的基础上。利用这几大模块的优点，能够大大提升机器学习的效率。
  sklearn拥有着完善的文档，上手容易，具备着丰富的API，在学术界颇受欢迎。sklearn已经封装了大量的机器学习算法，包括LIBSVM和LIBINEAR。同时sklearn内置了大量数据集，节省了获取和整理数据集的时间。

项目简介

链路预测是经过历史链接信息预测将来可能产生的链接，即经过当前网络中的连边信息预测未来可能产生的连边信息。

项目源码

from sklearn.model_selection import train_test_split # 分割数据模块
from sklearn.neighbors import KNeighborsClassifier # K最近邻(kNN，k-NearestNeighbor)分类算法
import pandas as pd
import numpy as np
from sklearn.model_selection import train_test_split
from sklearn.neighbors import KNeighborsClassifier
from sklearn import preprocessing
import matplotlib.pyplot as plt
from sklearn.svm import SVC
from math import isnan

定义计算共同邻居指标的方法

define some functions to calculate some baseline index

计算Jaccard类似性指标

def Jaccavrd(MatrixAdjacency_Train):

Matrix_similarity = np.dot(MatrixAdjacency_Train,MatrixAdjacency_Train)

deg_row = sum(MatrixAdjacency_Train)
deg_row.shape = (deg_row.shape[0],1)
deg_row_T = deg_row.T
tempdeg = deg_row + deg_row_T
temp = tempdeg - Matrix_similarity

Matrix_similarity = Matrix_similarity / temp
return Matrix_similarity

定义计算Salton指标的方法

def Salton_Cal(MatrixAdjacency_Train):

similarity = np.dot(MatrixAdjacency_Train,MatrixAdjacency_Train)

deg_row = sum(MatrixAdjacency_Train)
deg_row.shape = (deg_row.shape[0],1)
deg_row_T = deg_row.T
tempdeg = np.dot(deg_row,deg_row_T)
temp = np.sqrt(tempdeg)

np.seterr(divide='ignore', invalid='ignore')
Matrix_similarity = np.nan_to_num(similarity / temp)

Matrix_similarity = np.nan_to_num(Matrix_similarity)

return Matrix_similarity

def file2matrix(filepath):

f = open(filepath)
lines = f.readlines()
matrix = np.zeros((50, 50), dtype=float)
A_row = 0
for line in lines:
    list = line.strip('\n').split(' ')
    matrix[A_row:] = list[0:50]
    A_row += 1
return matrix

filepath = '3600/s0001.txt'
MatrixAdjacency = file2matrix(filepath)

similarity_matrix_Jaccavrd = Jaccavrd(MatrixAdjacency)
similarity_matrix_Salton = Salton_Cal(MatrixAdjacency)

filepath2 = '3600/s0002.txt'
MatrixAdjacency2 = file2matrix(filepath2)

similarity_matrix_Jaccavrd2 = Jaccavrd(MatrixAdjacency2)
similarity_matrix_Salton2 = Salton_Cal(MatrixAdjacency2)

filepath3 = '3600/s0003.txt'
MatrixAdjacency3 = file2matrix(filepath3)

similarity_matrix_Jaccavrd3 = Jaccavrd(MatrixAdjacency3)
similarity_matrix_Salton3 = Salton_Cal(MatrixAdjacency3)

获取jaccard类似性矩阵的行数和列数

Jaccard_Row = similarity_matrix_Jaccavrd.shape[0]
Jaccard_Column = similarity_matrix_Jaccavrd.shape[1]
Jaccard_List = []
for i in range(Jaccard_Row):

for j in range(Jaccard_Column):
    if i<j:
        index = similarity_matrix_Jaccavrd[i,j]
        if isnan(index) == True:
            index = 0
        Jaccard_List.append(index)

获取Salton类似性矩阵的行数和列数

Salton_Row = similarity_matrix_Salton.shape[0]
Salton_Column = similarity_matrix_Salton.shape[1]
Salton_List = []
for i in range(Salton_Row):

for j in range(Salton_Column):
    if i<j:
        index = similarity_matrix_Salton[i,j]
        if isnan(index) == True:
            index = 0
        Salton_List.append(index)

获取jaccard类似性矩阵的行数和列数

Jaccard_Row2 = similarity_matrix_Jaccavrd2.shape[0]
Jaccard_Column2 = similarity_matrix_Jaccavrd2.shape[1]
Jaccard_List2 = []
for i in range(Jaccard_Row2):

for j in range(Jaccard_Column2):
    if i<j:
        index2 = similarity_matrix_Jaccavrd2[i,j]
        if isnan(index2) == True:
            index2 = 0
        Jaccard_List2.append(index2)

获取Salton类似性矩阵的行数和列数

Salton_Row2 = similarity_matrix_Salton2.shape[0]
Salton_Column2 = similarity_matrix_Salton2.shape[1]
Salton_List2 = []
for i in range(Salton_Row2):

for j in range(Salton_Column2):
    if i<j:
        index2 = similarity_matrix_Salton2[i,j]
        if isnan(index2) == True:
            index2 = 0
        Salton_List2.append(index2)

获取jaccard类似性矩阵的行数和列数

Jaccard_Row3 = similarity_matrix_Jaccavrd3.shape[0]
Jaccard_Column3 = similarity_matrix_Jaccavrd3.shape[1]
Jaccard_List3 = []
for i in range(Jaccard_Row3):

for j in range(Jaccard_Column3):
    if i<j:
        index3 = similarity_matrix_Jaccavrd3[i,j]
        if isnan(index3) == True:
            index3 = 0
        Jaccard_List3.append(index3)

获取Salton类似性矩阵的行数和列数

Salton_Row3 = similarity_matrix_Salton3.shape[0]
Salton_Column3 = similarity_matrix_Salton3.shape[1]
Salton_List3 = []
for i in range(Salton_Row3):

for j in range(Salton_Column3):
    if i<j:
        index3 = similarity_matrix_Salton3[i,j]
        if isnan(index3) == True:
            index3 = 0
        Salton_List3.append(index3)

获取邻接矩阵的行数和列数

Adjacency_Row = MatrixAdjacency.shape[0]
Adjacency_Column = MatrixAdjacency.shape[1]
Adjacency = []
for i in range(Adjacency_Row):

for j in range(Adjacency_Column):
    if i<j:
        index = MatrixAdjacency[i,j]
        Adjacency.append(index)

获取邻接矩阵的行数和列数

Adjacency_Row2 = MatrixAdjacency2.shape[0]
Adjacency_Column2 = MatrixAdjacency2.shape[1]
Adjacency2 = []
for i in range(Adjacency_Row2):

for j in range(Adjacency_Column2):
    if i<j:
        index2 = MatrixAdjacency2[i,j]
        Adjacency2.append(index2)

获取邻接矩阵的行数和列数

Adjacency_Row3 = MatrixAdjacency3.shape[0]
Adjacency_Column3 = MatrixAdjacency3.shape[1]
Adjacency3 = []
for i in range(Adjacency_Row3):

for j in range(Adjacency_Column3):
    if i<j:
        index3 = MatrixAdjacency3[i,j]
        Adjacency3.append(index3)

data = np.zeros((1225,3))
data2 = np.zeros((1225,3))
data3 = np.zeros((1225,3))

for i in range(1225):

data[i][0] =  Jaccard_List[i]
data[i][1] = Salton_List[i]
data[i][2] = Adjacency[i]

for j in range(1225):

data2[j][0] =  Jaccard_List2[j]
data2[j][1] = Salton_List2[j]
data2[j][2] = Adjacency2[j]

for k in range(1225):

data3[k][0] =  Jaccard_List3[k]
data3[k][1] = Salton_List3[k]
data3[k][2] = Adjacency3[k]

data_train_X = data[:,0:2]
data_train_y = data[:,2]

data_test_X = data2[:,0:2]
data_test_y = data2[:,2]

data_target_X = data3[:,0:2]
data_target_y = data3[:,2]

knn = KNeighborsClassifier()
knn.fit(data_train_X,data_train_y)

print(knn.predict(data_test_X))

print(data_test_y)

clf = SVC()
clf.fit(data_train_X,data_test_y)

print(clf.score(data_test_X,data_target_y))

项目详细了解

如需详细本项目信息，可发送邮件至18770918982@gmail.com