回归中的相关度和R平方值——学习笔记

1. 相关度

相关度又叫 皮尔逊相关系数 (Pearson Correlation Coefficient):

衡量两个值线性相关强度的量 取值范围 [-1, 1]:  正向相关: >0, 负向相关：<0, 无相关性：=0

            cov：协方差  Var：方差

2. R平方值:

模型可以解释为多大程度是自变量导致因变量的改变

简单线性回归：R^2 = r * r

多元线性回归：

R平方也有其局限性：R平方随着自变量的增加会变大，R平方和样本量是有关系的。因此，我们要到R平方进行修正。得到R平方值adjusted，来评判线性回归模型的拟合度。修正的方法：

        

3. python的两种实现方法：

import numpy as np
# from astropy.units import  Ybarn
import math

def computeCorrelation(x, y):
    xBar = np.mean(x)
    ybar = np.mean(y)
    SSR = 0
    varX = 0
    varY = 0
    for i in range(0, len(x)): #多少实例
        diffxxBar = x[i] - xBar
        diffyyBar = y[i] - ybar
        SSR += (diffxxBar * diffyyBar)
        varX += diffxxBar ** 2      # 求平方然后累计起来
        varY += diffyyBar ** 2      # 求平方然后累计起来

    SST = math.sqrt(varX * varY)
    return SSR / SST

def polyfit(x, y, degree):

    result = {}     # 定义一个字典
    coeffs = np.polyfit(x, y, degree)   # 直接求出b0 b1 b2 b3 ..的估计值
    result["polynomial"] = coeffs.tolist()

    p = np.poly1d(coeffs)   # 返回预测值
    yhat = p(x)         # 传入x 返回预测值
    ybar = np.sum(y) / len (y)  # 求均值
    ssreg = np.sum((yhat -ybar)**2)
    sstot = np.sum((y - ybar)**2)
    result["determination"] = ssreg / sstot

    return result


testX = [1, 3, 8, 7, 9]
testY = [10, 12, 24, 21, 34]

print("r:", computeCorrelation(testX, testY))
print("r**2:", (computeCorrelation(testX, testY)**2))

print("r**2", polyfit(testX, testY, 1)["determination"])    # degree=1  一次
print(polyfit(testX, testY, 1)["polynomial"])       # 打印除斜率和截距