在Python中验证,只有线性回归时TSS=RSS+ESS,Ridge和Lasso时,TSS都大于RSS+ESS。
lr = LinearRegression().fit(x,y)ridge = Ridge(alpha=51).fit(x,y)lasso = Lasso().fit(x,y)models = [lr,ridge,lasso]for model in models:y_hat = model.predict(x)TSS = ((y-y_bar)**2).sum()ESS = ((y_hat-y_bar)**2).sum()RSS = ((y-y_hat)**2).sum()print('{:.4f}'.format(TSS-(ESS+RSS)))
-0.000093.60106245.2569
如果觉得《R^2(可决系数)为负分析》对你有帮助,请点赞、收藏,并留下你的观点哦!