Add files via upload

5주차/LinearRegression_조원국.cpp

@@ -0,0 +1,150 @@
+#include <iostream>
+#include <vector>
+#include <numeric>
+#include <cmath>
+#include <random>
+
+using namespace std;
+
+class LinearRegression
+{
+private:
+    double mCoef;
+    double mIntercept;
+    double mRsquared;
+    unsigned long mSize;
+    double mMean_x;
+    double mMean_y;
+
+public:
+    void Fit(vector<double> tTrain_x, vector<double> tTrain_y)
+    {
+        mSize = tTrain_x.size();
+        mMean_x = std::accumulate(tTrain_x.begin(), tTrain_x.end(), 0.0) / mSize;
+        mMean_y = std::accumulate(tTrain_y.begin(), tTrain_y.end(), 0.0) / mSize;
+
+        double temp_nr = 0.0;
+        double temp_dr = 0.0;
+        for (unsigned i = 0; i < mSize; i++)
+        {
+            temp_nr += (tTrain_x[i] - mMean_x) * (tTrain_y[i] - mMean_y);
+            temp_dr += pow( (tTrain_x[i] - mMean_x), 2);
+        }
+
+        mCoef = temp_nr / temp_dr;
+        mIntercept = mMean_y - mCoef * mMean_x;
+
+        double rss = 0.0;
+        double tss = 0.0;
+        for (unsigned i = 0; i < mSize; i++)
+        {
+        rss += pow( (tTrain_y[i] - (tTrain_x[i] * mCoef + mIntercept)), 2);
+        tss += pow( (tTrain_x[i] - mMean_x), 2);
+        }
+
+        mRsquared = 1.0 - (rss / tss);
+    }
+
+    void GetParams()
+    {
+        cout << "Coefficient = " << mCoef << ", Intercept = " << mIntercept << endl;
+    }
+
+    void GetScore()
+    {
+        cout << "R-squared = " << mRsquared << endl;
+    }
+
+    void Predict(vector<double> tTest_x, vector<double> tTest_y)
+    {
+        double mse = 0.0;
+        for (unsigned i = 0; i < mSize; i++)
+        {
+            mse += pow( ( tTest_y[i] - (tTest_x[i] * mCoef + mIntercept) ), 2);
+        }
+        double rmse = sqrt(mse / mSize);
+        cout << "RMSE = " << rmse << endl;
+    }
+};
+
+class DataGenerator
+{
+public:
+    static double Coef;
+    static double Intercept;
+    static double Std;
+    static int Size;
+    static int Min;
+    static int Max;
+    vector<double> Vector_x;
+    vector<double> Vector_y;
+
+private:
+    double mRandValue;
+    double mRandNoise;
+
+public:
+    void Generate()
+    {
+        std::random_device rand_seed;
+        std::mt19937 generator(rand_seed());
+        std::uniform_int_distribution<double> uniform(Min, Max);
+        std::normal_distribution<double> normal(0, Std);
+
+        for (unsigned i = 0; i < Size; i++)
+        {
+            mRandValue = uniform(generator);
+            mRandNoise = normal(generator);
+            Vector_x.push_back(mRandValue);
+            Vector_y.push_back(mRandValue * Coef + Intercept + mRandNoise);
+        }
+    }
+
+    vector<double> SampleX()
+    {
+        cout << '[';
+        for (unsigned i = 0; i < 20; i++)
+        {
+            cout << Vector_x[i] << ", ";
+        }
+        cout << ']' << endl;
+        return Vector_x;
+    }
+
+    vector<double> SampleY()
+    {
+        cout << '[';
+        for (unsigned i = 0; i < 20; i++)
+        {
+            cout << Vector_y[i] << ", ";
+        }
+        cout << ']' << endl;
+        return Vector_y;
+    }
+};
+
+double DataGenerator::Coef = -5;
+double DataGenerator::Intercept = 10;
+double DataGenerator::Std = 10;
+int DataGenerator::Min = 0;
+int DataGenerator::Max = 1000;
+int DataGenerator::Size = 100000;
+
+int main()
+{
+    DataGenerator DataTrain;
+    DataTrain.Generate();
+    vector<double> dataTrain_x = DataTrain.SampleX();
+    vector<double> dataTrain_y = DataTrain.SampleY();
+
+    DataGenerator DataTest;
+    DataTest.Generate();
+    vector<double> dataTest_x = DataTest.SampleX();
+    vector<double> dataTest_y = DataTest.SampleY();
+
+    LinearRegression LinearModel;
+    LinearModel.Fit(dataTrain_x, dataTrain_y);
+    LinearModel.GetParams();
+    LinearModel.GetScore();
+    LinearModel.Predict(dataTest_x, dataTest_y);
+}