/*
 * @Author: SongZihui-sudo [email protected]
 * @Date: 2024-02-24 14:48:58
 * @LastEditors: SongZihui-sudo [email protected]
 * @LastEditTime: 2024-02-24 15:44:26
 * @FilePath: /linear_regression/src/linear_regression.h
 * @Description: 这是默认设置,请设置`customMade`, 打开koroFileHeader查看配置 进行设置:
 * https://github.com/OBKoro1/koro1FileHeader/wiki/%E9%85%8D%E7%BD%AE
 */
#pragma once

#include <tuple>	
#include <vector>

class linear_regression
{
public:
    linear_regression( const std::vector< std::tuple< double, double > > trains = { },
                       const std::vector< std::tuple< double, double, double > > tests = { },
                       const double leaning_rate = 0.5 )
    {
        alapa       = leaning_rate;
        m_train_set = trains;
        m_test_set  = tests;
    }
    ~linear_regression( ){ };

public:
    void train( );

    void test( );

    double test( const double x );

    double cost( );

    double get_w( ) { return w; }

    double get_b( ) { return b; }

    void set_train_set( const std::vector< std::tuple< double, double > > list )
    {
        m_train_set = list;
    }

    void set_test_set( const std::vector< std::tuple< double, double, double > > list )
    {
        m_test_set = list;
    }

private:
    double w_diff( );

    double b_diff( );

    double f_w_b( const double x ) { return w * x + b; }

private:
    std::vector< std::tuple< double, double > > m_train_set;
    std::vector< std::tuple< double, double, double > > m_test_set;
    double w = 0;
    double b = 0;
    double alapa;
};
        

    
/*
 * @Author: SongZihui-sudo [email protected]
 * @Date: 2024-02-24 14:48:58
 * @LastEditors: SongZihui-sudo [email protected]
 * @LastEditTime: 2024-02-24 16:21:10
 * @FilePath: /linear_regression/src/linear_regression.c
 * @Description: 这是默认设置,请设置`customMade`, 打开koroFileHeader查看配置 进行设置:
 * https://github.com/OBKoro1/koro1FileHeader/wiki/%E9%85%8D%E7%BD%AE
 */
#include "linear_regression.h"

#include <cmath>
#include <<iostream>

void linear_regression::train( )
{
    while ( w_diff( ) && b_diff( ) )
    {
        double temp_w    = w - alapa * w_diff( );
        double temp_b    = b - alapa * b_diff( );
        w                = temp_w;
        b                = temp_b;
        double temp_cost = cost( );
        std::cout << "Current Cost: " << temp_cost << std::endl;
        if ( std::isnan( temp_cost ) || std::isinf( temp_cost ) )
        {
            break;
        }
    }
    std::cout << "Train Complete!" << std::endl;
    std::cout << "Y = " << w << "* x + (" << b << ")" << std::endl;
}

double linear_regression::test( const double x )
{
    double y = f_w_b( x );
    return y;
}

void linear_regression::test( )
{
    for ( std::size_t i = 0; i < m_test_set.size( ); i++ )
    {
        double x                       = std::get< 0 >( m_test_set[i] );
        double y                       = f_w_b( x );
        std::get< 2 >( m_test_set[i] ) = y;
        std::cout << "Y(" << x << ") = " << y << std::endl;
    }
}

double linear_regression::cost( )
{
    double sum = 0;
    for ( std::size_t i = 0; i < m_train_set.size( ); i++ )
    {
        double x = std::get< 0 >( m_train_set[i] );
        double y = std::get< 1 >( m_train_set[i] );
        sum += std::pow( f_w_b( x ) - y, 2 );
    }
    return sum / ( 2 * m_train_set.size( ) );
}

double linear_regression::w_diff( )
{
    double sum = 0;
    for ( std::size_t i = 0; i < m_train_set.size( ); i++ )
    {
        double x = std::get< 0 >( m_train_set[i] );
        double y = std::get< 1 >( m_train_set[i] );
        sum += ( f_w_b( x ) - y ) * x;
    }
    return sum / m_train_set.size( );
}

double linear_regression::b_diff( )
{
    double sum = 0;
    for ( std::size_t i = 0; i < m_train_set.size( ); i++ )
    {
        double x = std::get< 0 >( m_train_set[i] );
        double y = std::get< 1 >( m_train_set[i] );
        sum += ( f_w_b( x ) - y );
    }
    return sum / m_train_set.size( );
}