tidalmelon / lr

线性回归， 逻辑回归

1 linear regression and logistic regression

1, 一元二次方程求极值

梯度下降法: lr_1.cpp lr_1.py

while ( max_iterate_num or other ) {
    x = x - alpha * 导数(f(x))
}

2, 一元线性回归

g++ lr_2.cpp
After 1500 iterates, the cost Error(w0, w1) is *41.844789*
w0 = [0.018653], w1 = [2.981086]
predict(112) = 333.900241
predict(110) = 327.938070

2.1 最小二乘拟合

python least_square.py

least square error cost(-23.5512952764, 3.20708095323) is *35.1218735092*
predict(112) =  335.641771485
predict(110) =  329.227609578

2.2 一元线性回归 与 最小二乘拟合对比

3, 多元线性回归

g++ lr_3.cpp
After 1500 iterates, the cost Error(w0, w1) is 40.559510
w0=-0.053241
w1=2.973633
w2=0.364542
w3=0.196225
w4=-0.198871
predict(112) = 325.488320
predict(110) = 325.977873

4, 逻辑回归

sigmoid 函数：