机器学习算法自适应增强AdaBoost之c++实现（不调用外源库）

本文链接：https://blog.csdn.net/2401_84975241/article/details/138982669

既有适合小白学习的零基础资料，也有适合3年以上经验的小伙伴深入学习提升的进阶课程，涵盖了95%以上C C++开发知识点，真正体系化！

由于文件比较多，这里只是将部分目录截图出来，全套包含大厂面经、学习笔记、源码讲义、实战项目、大纲路线、讲解视频，并且后续会持续更新

#include
#include <math.h>
#include
#include
#include
#include “matrix.h”

using namespace std;

#define MAX 1000000
#define MIN -100000

struct Stump {
map<string, int>stumpDim;
map<string, double>stumpThresh;
map<string, string>stumpIneq;
map<string, double>stumpAlpha;
};

vectorweakClassArr;

//求每列的最大最小
void rangeSize(Matrix &Data, int &axis, vector &rangeVal)
{
rangeVal.clear();
rangeVal.push_back(MIN);
rangeVal.push_back(MAX);
for (int i = 0; i < Data.col; i++)
{
if (Data.mat[i][axis]>rangeVal[0])
rangeVal[0] = Data.mat[i][axis];
if (Data.mat[i][axis]<rangeVal[1])
rangeVal[1] = Data.mat[i][axis];
}
}

vector stumpClassify(Matrix Data, int axis, double threshVal, string threshIneq)
{
vector retArray(Data.col, 1);
for (int i = 0; i < Data.col; i++)
{
if (threshIneq == “lt”)
{
if (Data.mat[i][axis] <= threshVal)
{
retArray[i] = -1;
}
}
else
{
if (Data.mat[i][axis] > threshVal)
{
retArray[i] = -1;
}
}
}
return retArray;
}

//return 最小的误差
double buildStump(Matrix &dataMatrix, Matrix &labelMat, Stump &bestStump, vector &bestClasEst, vector &D)
{
int m = dataMatrix.col;//行
int n = dataMatrix.row;//列
double numSteps = 10.0;
vectorrangeVal;
vectorpredictedVals;
vectorthreshIneq{ “lt”,“gt” };
double minError = INT_MAX;
for (int i = 0; i < n; i++)
{
rangeSize(dataMatrix, i, rangeVal);
double rangeMin = rangeVal[1];
double rangeMax = rangeVal[0];
double stepSize = (rangeMax - rangeMin) / numSteps;
for (int j = -1; j < int(numSteps) + 1; j++)
{
for (int k = 0; k < 2; k++)
{
double threshVal = rangeMin + j * stepSize;
predictedVals = stumpClassify(dataMatrix, i, threshVal, threshIneq[k]);
vectorerrArr(m, 1);
for (int ii = 0; ii < m; ii++)
{
if (predictedVals[ii] == labelMat.mat[ii][0])
{
errArr[ii] = 0;
}
}
double weightedError = 0.0;
for (int ii = 0; ii < m; ii++)
{
weightedError += D[ii] * errArr[ii];
}
// cout<<“split: dim “<<i<<”,thresh “<<threshVal<<”,thresh inequal:”<<threshIneq[k];
// cout<<", the weighted error is "<<weightedError<<endl;
if (weightedError < minError)
{
minError = weightedError;
bestClasEst = predictedVals;
bestStump.stumpDim[“dim”] = i;
bestStump.stumpThresh[“thresh”] = threshVal;
bestStump.stumpIneq[“ineq”] = threshIneq[k];
}
}
}
}
return minError;
}