python - 返回 beta 系数 B0 和 B1 的函数

Question

这是我想要做的。创建一个采用两个系列 (x,y) 的函数，然后计算并返回最小化 SSE 的截距和斜率系数。

我在下面编写了一个函数，它采用两个系列并返回线性回归方程的截距（B0）和斜率（B1）。问题？这些价值观在我看来是错误的。我知道这是编写此函数的一种粗略方法。关于如何改进的提示？

import pandas as pd
import numpy as np

def coeff(Series_y,Series_x):
    beta_0 = 0
    beta_1 = 0
    y_list = Series_y.tolist()
    x_list = Series_x.tolist()
    y_list_array = np.array(y_list)
    x_list_array = np.array(x_list)
    y_list_mean = np.mean(y_list_array)
    x_list_mean = np.mean(x_list_array)
    y_list_stdev = np.std(y_list_array)
    x_list_stdev = np.std(x_list_array)
    r_x_y = np.corrcoef(x_list_array,y_list_array)
    beta_1 = r_x_y * (y_list_stdev/x_list_stdev)
    beta_0 = y_list_mean - (beta_1 * x_list_mean)
    return beta_0
    return beta_1
    print(beta_0, beta_1)

我重写了这个函数。它仍然非常笨重。

# create the needed variables
beta_0 = 0
beta_1 = 0
r_x_y = 0
y_list = 0
x_list = 0
y_list_array = 0
x_list_array = 0
y_list_mean = 0
x_list_mean = 0
y_list_stdev = 0
x_list_stdev = 0 
def coeff(Series_y,Series_x):
    # create the two needed beta variables
    # calculate needed stats
    y_list = Series_y.tolist()
    x_list = Series_x.tolist()
    y_list_array = np.array(y_list)
    x_list_array = np.array(x_list)
    y_list_mean = np.mean(y_list_array)
    x_list_mean = np.mean(x_list_array)
    y_list_stdev = np.std(y_list_array)
    x_list_stdev = np.std(x_list_array)    
    r_x_y = np.corrcoef(x_list_array,y_list_array)
    return x_list_mean, y_list_mean, x_list_stdev, y_list_stdev, r_x_y
    # calculate the beta coefficients
    beta_1 = r_x_y * (y_list_stdev/x_list_stdev)
    beta_0 = y_list_mean - (beta_1 * x_list_mean)
    return beta_0, beta_1
    # Return line of best fit.
    return beta_0 + beta_1 * x
    print(beta_0 + beta_1 * x)    

我使用比尔的建议重写了它。容易多了！

def coeff(Series1, Series2):
    return print(scipy.stats.linregress(Series1, Series2))

标签： pythonfunctionstatistics