gistfile1.txt

def regression_analysis(datas):
    # x, yの平均値を求める
    x_ave = sum([d[0] for d in datas]) / len(datas)
    y_ave = sum([d[1] for d in datas]) / len(datas)
    
    # x, yの分散、xyの共分散を求める
    c_xy = sum([(d[0] - x_ave) * (d[1] - y_ave) for d in datas]) / len(datas)
    d_x = sum([(d[0] - x_ave) ** 2 for d in datas]) / len(datas)
    d_y = sum([(d[1] - y_ave) ** 2 for d in datas]) / len(datas)
    
    # 回帰方程式のパラメータa, bを求める
    a = (c_xy / d_x)
    b = (y_ave - (c_xy / d_x) * x_ave)
    print('y = {a}x + {b}'.format(a=a, b=b))
    
    # yの予測値hat_yの分散を求めたら決定係数を求める
    hat_y = [a * d[0] + b for d in datas]
    d_hat_y = sum([(hy - y_ave) ** 2 for hy in hat_y]) / len(datas)
    r2 = d_hat_y / d_y
    print('R^2 = {r2}'.format(r2=r2))