学习笔记第十篇之安全评估模型设计

        最近老师的项目需要做一个安全评估模型，想到以后可能会用到，所以在此也记录一下。主要是根据已知标签的数据来预测后面数据的标签值。这里的标签是0或1，相当于二分类问题，所以我这里使用神经网络来训练模型，用的是python中的sklearn这个包。数据如下：

3,4,7,4,5,2,0,1,1
1,3,7,4,5,2,0,0.3,0
6,4,5,4,5,2,0,1,1
5,3,5,4,5,2,0,0.3,0
5,4,7,7,5,2,0,0.5,1
6,4,5,7,5,2,0,0.5,0
4,8,7,4,7,2,75,0.5,1
3,8,6,4,2,1,0,1,1
3,8,6,4,2,1,0,1,1
3,8,6,4,2,1,0,1,0
8,4,7,7,5,1,0,0.2,1
9,3,5,9,3,1,0,0.01,0
9,8,6,4,2,1,0,1,1
9,3,5,9,3,1,0,0.01,0
9,8,6,4,2,1,0,1,1
9,8,6,8,4,2,0,0.11,0
9,8,5,8,4,2,0,0.11,1
5,6,7,8,5,2,0,0.2,1
6,6,7,8,5,2,0,0.2,0
4,4,5,4,5,1,0,0.11,0
3,8,6,4,2,1,0,1,1
3,8,6,4,2,1,0,1,1
3,8,7,4,7,2,75,0.5,1

当然，这里数据每个属性的值是由具体的物理意义的，实际使用中的数据非常多。代码如下：

# coding: utf-8

# 导入需要的包
from sklearn.neural_network import MLPClassifier
import pandas as pd
import numpy as np

#获取数据函数，返回数据属性列表和标签列表
def read_data():
    f = open('rfid_data.csv')        #读取文件数据,路径可能需要修改
    df = pd.read_csv(f)
    data_x = df.iloc[:, :8].values
    data_y = df.iloc[:,8].values
    normalized_data = (data_x - np.mean(data_x, axis=0))/np.std(data_x,axis=0)
    x_list = []
    y_list = []
    for i in range(len(normalized_data)):
        x = normalized_data[i,:]
        y = data_y[i]
        x_list.append(list(x))
        y_list.append(y)
    return x_list, y_list

#结果评估函数
def evaluate(array):
    system_score = np.mean(array)
    if system_score <= 0.1:
        print 'the system is safe.'
    elif 0.1 <system_score <= 0.2:
        print 'the system is safer.'
    elif 0.2 <system_score <= 0.5:
        print 'the system is medium security.'
    elif 0.5 <system_score <= 0.8:
        print 'the system is low security.'
    elif system_score > 0.8:
        print 'the system is insecurity.'

if __name__ == '__main__':
    X, y = read_data()
    train_x = X[0:20]
    train_y = y[0:20]
    test_x = X[20:]
    test_y = y[20:]
    # 运用神经网络模型训练
    clf = MLPClassifier(solver='lbfgs', alpha=1e-5,hidden_layer_sizes=(5, 1), random_state=1)
    clf.fit(train_x,train_y)
    prediction = clf.predict(test_x)
    evaluate(prediction)