运动状态的程序编写
# -*- coding: utf-8 -*-
"""
Created on Sun Apr 15 14:20:53 2018
@author: Administrator
"""
import pandas as pd
import numpy as np
# 从sklearn库中导入预处理模块 Imputer
from sklearn.preprocessing import Imputer
# 导入自动生成训练集和测试集的模块 train_test_split
from sklearn.cross_validation import train_test_split
# 导入预测结果评估模块 classification_report
from sklearn.metrics import classification_report
# 导入K近邻分类器 决策树分类器 高斯朴素贝叶斯
from sklearn.neighbors import KNeighborsClassifier
from sklearn.tree import DecisionTreeClassifier
from sklearn.naive_bayes import GaussianNB
# 读取特征文件列表和标签文件中的内容,归并后返回
def load_datasets(feature_paths, label_paths):
feature = np.ndarray(shape=(0,41))
label = np.ndarray(shape=(0,1))
for file in feature_paths:
# 使用pandas库的read_table函数读取一个特征文件内容
# 指定分隔符为逗号 缺失值为问号 文件中不包含表头行
df = pd.read_table(file, delimiter=',', na_values='?', header=None)
# 使用Imputer函数,通过设定strategy参数为'mean'
# 使用平均值对缺失数据补全,fit()函数用于训练预处理器,
# transform()函数用于生成预处理结果
imp = Imputer(missing_values='NaN', strategy='mean', axis=0)
imp.fit(df)
df = imp.transform(df)
# 将预处理后的数据加入feature,依次遍历完所有特征文件
feature = np.concatenate((feature, df))
for file in label_paths:
# 同上
df = pd.read_table(file, header=None)
# 标签文件没有缺失值,所以直接将读取到的新数据加入label集合
label = np.concatenate((label, df))
label = np.ravel(label)
# 将特征集合feature与标签集合label返回
return feature, label
if __name__ == '__main__':
''' 数据路径 '''
featurePaths = ['A/A.feature','B/B.feature','C/C.feature']
labelPaths = ['A/A.label','B/B.label','C/C.label']
''' 读入数据 '''
# 使用分片方法,将数据路径中前2个数据作为训练集传入load_datasets()
# 得到训练集的特征x_train,训练集的标签y_train
x_train,y_train = load_datasets(featurePaths[:2],labelPaths[:2])
# 将最后一个数据作为测试集,传入load_datasets()
# 得到测试集的特征x_test,训练集的标签y_test
x_test,y_test = load_datasets(featurePaths[2:],labelPaths[2:])
# 使用train_test_split()函数,通过设置测试集比例test_size为0,将数据随机打乱,便于后续分类器的初始化和训练
x_train, x_, y_train, y_ = train_test_split(x_train, y_train, test_size = 0.0)
# 创建K近邻分类器,并将训练集x_train和y_train传入fit()函数进行训练
print('Start training knn')
knn = KNeighborsClassifier().fit(x_train, y_train)
print('Training done')
# 使用测试集x_test,进行分类器预测,得到分类结果
answer_knn = knn.predict(x_test)
print('Prediction done')
# 创建决策树分类器,并将训练集x_train和y_train传入fit()函数进行训练
print('Start training DT')
dt = DecisionTreeClassifier().fit(x_train, y_train)
print('Training done')
answer_dt = dt.predict(x_test)
print('Prediction done')
# 创建贝叶斯分类器,并将训练集x_train和y_train传入fit()函数进行训练
print('Start training Bayes')
gnb = GaussianNB().fit(x_train, y_train)
print('Training done')
answer_gnb = gnb.predict(x_test)
print('Prediction done')
# 使用classification_report()函数对分类结果,从精确率precision 召回率recall f1值f1-sorce和支持度support四个维度进行衡量
# 分别对三个分类结果进行输出
print('\n\nThe classification report for knn:')
print(classification_report(y_test, answer_knn))
print('\n\nThe classification report for DT:')
print(classification_report(y_test, answer_dt))
print('\n\nThe classification report for Bayes:')
print(classification_report(y_test, answer_gnb))
程序运行结果:
结论:
从准确度的角度衡量,贝叶斯分类器的效果最好
从召回率和F1值得角度衡量,k近邻效果最好
贝叶斯分类器和k近邻效果好于决策树