PCATest.cpp

Go to the documentation of this file.

//
// Created by Xinyu Zhang on 3/26/21.
//
// gcc -I/Users/xinyuzhang/Desktop/Spring2021/c++pattern/project/cosan /Users/xinyuzhang/Desktop/Spring2021/c++pattern/project/cosan/test/preprocessor/PCATest.cpp 
 
#include <iostream>
#include <Eigen/Dense>
#include <Eigen/Core>
#include <cosan/data/CosanData.h>
#include <gsl/gsl>
#include <cosan/preprocessing/overunderflow.h>
#include <cosan/preprocessing/missingvalues.h>
#include <cosan/preprocessing/minmaxscaler.h>
#include <cosan/preprocessing/principalcomponentanalysis.h>
 
#include <vector>
#ifndef FMT_HEADER_ONLY
#define FMT_HEADER_ONLY
#endif
#include <fmt/format.h>
#include <numeric>
//using namespace Eigen;
//using namespace std;
 
 
int main() {
 
    // std::cout<<  value1<<value2<<std::endl;
    // std::cout<<X.col(i).array().isNaN().sum()<<std::endl;
    Cosan::CosanMatrix X{5,3};
    X << 1,2,2,
         4,3,4,
         2,4,6,
         12,51,6,
         435,4,15;               
    // (X.array().colwise()-X.colwise().minCoeff())/X.colwise().maxCoeff()-X.colwise().minCoeff();
    std::cout<<X<<std::endl;
 
    Cosan::CosanMatrix centered = X.rowwise() - X.colwise().mean();
    Cosan::CosanMatrix cov = centered.adjoint() * centered;
 
    Eigen::SelfAdjointEigenSolver<Cosan::CosanMatrix> eigensolver(cov);
    if (eigensolver.info() != Eigen::Success) {
        throw std::invalid_argument("Cannot solve eigenvalue decomposition.");}
    std::cout<<eigensolver.eigenvalues()<<std::endl;
    std::cout<<eigensolver.eigenvectors()<<std::endl;
    std::cout<<eigensolver.eigenvectors().rightCols(2)<<std::endl;  
    centered = eigensolver.eigenvectors().rightCols(2);
    std::vector<double> vec(eigensolver.eigenvalues().data(), eigensolver.eigenvalues().data() + eigensolver.eigenvalues().size());
    std::reverse(vec.begin(),vec.end());
    std::vector<double> runningSum(vec.size());
    std::partial_sum(vec.begin(), vec.end(), runningSum.begin(), std::plus<double>());
    for (auto & each :runningSum){
        std::cout<<each<<std::endl;
    }
 
 
 
    Cosan::CosanRawData CRD("./example_data/toy2/X_.csv","./example_data/toy2/Y_.csv");
    std::cout<<CRD.GetSummaryMessageX()<<CRD.GetSummaryMessageY()<<std::endl;
    Cosan::OverUnderFlow ouf(CRD);
    Cosan::MissingValues MissV(CRD);
    Cosan::MinmaxScaler MMS(CRD,3,5);       
    auto [value1, value2] = CRD.GetMissingNumber();
    std::cout<< value1<<value2<<std::endl;
    Cosan::PrincipalComponentAnalysis PCA(CRD,4);
 
    // std::cout<<M.array().isNaN().sum()<<std::endl;
    // CRD.GetTarget().rows()-CRD.GetTarget().array().isNaN().sum()
    // CRD.GetTarget().array().isNaN().select(0,CRD.GetTarget())
 
    // std::cout<<CRD.GetTarget().array().isNaN().select(0,CRD.GetTarget()).sum()<<std::endl;
 
    // std::cout<<(1-CRD.GetTarget().array().isNaN()).cast<double>().array()*CRD.GetTarget().array()<<std::endl;
 
 
    return 0;
}