人工智能（AI）之KNN的基本实现
数据集下载地址点我下载
本文主要介绍KNN的实现思想：
- KNN的主要思想就是：通过计算训练集与测试集之间的距离（欧氏距离、余弦距离、曼哈顿距离等），然后取出最相似的前N个数据对测试集进行预测
- 通过测试之后发现，就本次的数据集而言，把余弦距离以及欧氏距离进行加权来确定预测值结果较好，但仅仅是对于本次的训练数据而言
- KNN当中也还有很多细节可以去优化的，比如说对数据集进行一定的归一化，而归一化的方法也是很多的，具体怎么取，也是要看当前的数据集，找到适合的才是最好的
- 总之对于预测，找好模型才是最重要的，框架确定之后，再来讨论具体的优化会更有效果
#include
else
{
ss >> str;
sets.insert(str);
}
}
}
}
}
}else{
cerr<<"open in or out file error"<for (set<string>::iterator it = sets.begin();it != sets.end();it++)
{
if(*it != " ")
{
out << *it << endl;
//cout << *it << endl;
}
}
in.close();
out.close();
}
void clear_stopwords()
{
fstream in;
in.open("G:/桌面文档/学/大三上学期/第二学期/人工智能/实验/Lab 2 实验材料/Foxstoplist (1).txt");
ofstream out("G:/桌面文档/学/大三上学期/第二学期/人工智能/实验/Lab 2 实验材料/Foxstoplistout.txt");
string str;
if(in)
{
while(getline(in,str))
{
stringstream ss;
ss << str;
while(!ss.eof())
{
ss >> str;
out << str <for(set<string>::iterator it = sets.begin();it != sets.end();)
{
if(*it == str)
{
sets.erase(it);
break;
}
else
{
it++;
}
}
}
}
}
in.close();
out.close();
}
void vector_out()
{
ifstream in("G:/桌面文档/学/大三上学期/第二学期/人工智能/实验/Lab 2 实验材料/Dataset_words.txt");
ofstream out("G:/桌面文档/学/大三上学期/第二学期/人工智能/实验/Lab 2 实验材料/vector.txt");
string str;
int i = 0;
int row_num = 0;
while(in&&out)
{
while(getline(in,str))
{
if(i==0)
{
i++;
continue;
}
else
{
int j = 0;
stringstream ss;
ss << str;
while(!ss.eof())
{
int lin_num = 0;
if(j==0)
{
j++;
ss >> str;
}
else
{
ss >> str;
for(set<string>::iterator it=sets.begin(); it != sets.end() ; it++)
{
if(*it == str)
{
vector_old[row_num][lin_num] = true;
}
lin_num++;
}
}
}
}
row_num++;
}
}
string wenben = "文本编号 ";
out << wenben;
for(set<string>::iterator it= sets.begin(); it != sets.end(); it++)
{
out << *it << " ";
}
in.close();
out.close();
}
void compute_dis(double K)
{
for (int i = 0; i < 1246; i++){
double sum = 0;
for (int j = 0; j < sets.size(); j++){
if (vector_old[i][j])
{
sum++;
}
}
for (int j = 0; j < sets.size(); j++){
vector2[i][j] = vector_old[i][j]*1.0/sum;
//out << vector2[i][j] << " ";
}
//out <
}
for(int mood_n = 0 ; mood_n < 6 ; mood_n++)
{
for(int i = 0 ; i < 1000 ; i++)
{
int dis_num=0;
double pro_sum = 0;
double dis;
int pos;
double max_dis = 0;
double min_dis = 10000;
map<double,int>map1;
map<double,int, greater<double> >map2;
for(int j = 0 ; j < 246 ; j++)
{
dis = 0;
double angle = 0;
double xy_sum=0;
double xx=0;
double yy=0;
double save_angle[2000]={0};
for(int k = 0 ; k < sets.size() ; k++)
{
xy_sum+=vector_old[i+246][k]*vector_old[j][k];
xx+=vector_old[i+246][k]*vector_old[i+246][k];
yy+=vector_old[j][k]*vector_old[j][k];
//dis += (vector2[i+246][k]-vector2[j][k])*(vector2[i+246][k]-vector2[j][k]);
}
dis_save[j] = xx + yy - 2*xy_sum;
angle = xy_sum/(sqrt(xx)*sqrt(yy));
//angle = angle*(1/sqrt(dis_save[j]));
angle = 0.8*angle + 0.2*dis_save[j];
map2.insert(make_pair(angle,j));
/*
for(int k = 0 ; k < sets.size() ; k++)
{
dis += (vector2[i+246][k]-vector2[j][k])*(vector2[i+246][k]-vector2[j][k]);
}
dis = sqrt(dis);
dis_sum+=dis;
map1.insert(make_pair(dis,j));
*/
}
cout << "i:" << i </*
for(map::iterator it1 = map1.begin();it1!=map1.end();it1++)
{
double temp = it1->first;
temp = temp/dis_sum;
map1.insert(make_pair(temp,it1->second));
}
*/
int K_i = 1;
double dis_sum = 0;
for(map<double,int>::iterator it = map2.begin(); it != map2.end(); it++)
{
if(K_i>K)
{
break;
}
else
{
K_i++;
pro_sum += proba[mood_n][it->second];
//dis_sum+=(1/(dis_save[it->second]*dis_save[it->second]));
/*
for(map::iterator it1 = map1.begin();it1!=map1.end();it1++)
{
if(it->second == it1->second)
{
pro_sum = pro_sum + 0.6*
break;
}
}
*/
}
}
newproba[mood_n][i] = pro_sum*1.0/K;
}
}
cout << "happy" <void print()
{
for(int i = 0 ; i < 6 ; i++)
{
ofstream f;
switch(i)
{
case ANGER: f.open("C:/Users/windowos 7/Desktop/AILab/predict_test/anger_predict.txt"); break;
case DISGUST: f.open("C:/Users/windowos 7/Desktop/AILab/predict_test/disgust_predict.txt"); break;
case FEAR: f.open("C:/Users/windowos 7/Desktop/AILab/predict_test/fear_predict.txt"); break;
case JOY: f.open("C:/Users/windowos 7/Desktop/AILab/predict_test/joy_predict.txt"); break;
case SAD: f.open("C:/Users/windowos 7/Desktop/AILab/predict_test/sad_predict.txt"); break;
case SURPRISE: f.open("C:/Users/windowos 7/Desktop/AILab/predict_test/surprise_predict.txt"); break;
}
for(int j = 0 ; j < 1000 ; j++)
{
f << newproba[i][j] <//cout << newproba[i][j] <
}
f.close();
}
}
int main()
{
cout << "请输入k：" <cin >> K;
get_word();
cout << 0 <cout << 1 <cout << 2 <cout << 3 <cout << 4 <cout << 5 <cout << sets.size() <return 0;
}