如何解释MANE_TEST_SCORE和STD_TEST_SCORE？
我正在尝试使用具有分数=“ R2”的GridSearchCV来调整角rregressor模型的Hyperparams。 代码
means = gs.cv_results_['mean_test_score']
stds = gs.cv_results_['std_test_score']
params = gs.cv_results_['params']
for mean, stdev, param in zip(means, stds, params):
print("Summary: %f (%f) with: %r" % (mean, stdev, param))
输出以下结果。我担心R2是如此不同。特别是负面因素，意味着该模型的性能差。 这是否意味着存在潜在的问题？
Summary: -0.005585 (0.234004) with: {'batch_size': 5,}
Summary: 0.849829 (0.023618) with: {'batch_size': 5, batch_size': 10, 'epochs': 100}
Summary: -0.210410 (0.131035) with: {'batch_size': 20, 'epochs': 10}
Summary: 0.294107 (0.261391) with: {'batch_size': 20, 'epochs': 50}
Summary: 0.857905 (0.006955) with: {'batch_size': 20, 'epochs': 100}
Summary: -0.381018 (0.118675) with: {'batch_size': 40, 'epochs': 10}
Summary: -0.053815 (0.321491) with: {'batch_size': 40, 'epochs': 50}
Summary: 0.582999 (0.195620) with: {'batch_size': 40, 'epochs': 100}
Summary: -0.413765 (0.114058) with: {'batch_size': 60, 'epochs': 10}
Summary: -0.018707 (0.090325) with: {'batch_size': 60, 'epochs': 50}
Summary: 0.205208 (0.157954) with: {'batch_size': 60, 'epochs': 100}
Summary: -0.423384 (0.112643) with: {'batch_size': 80, 'epochs': 10}
Summary: -0.077780 (0.101668) with: {'batch_size': 80, 'epochs': 50}
Summary: -0.000941 (0.345382) with: {'batch_size': 80, 'epochs': 100}
Summary: -0.427151 (0.112105) with: {'batch_size': 100, 'epochs': 10}
Summary: -0.177046 (0.125664) with: {'batch_size': 100, 'epochs': 50}
培训数据集形状为（9073，5），测试：（477，5）
谢谢
I'm trying to tune hyperparams of the KerasRegressor model using GridSearchCV that has score = "r2" .
The code
means = gs.cv_results_['mean_test_score']
stds = gs.cv_results_['std_test_score']
params = gs.cv_results_['params']
for mean, stdev, param in zip(means, stds, params):
print("Summary: %f (%f) with: %r" % (mean, stdev, param))
outputs the results below. I'm concerned that r2 is so different. Especially negatives that mean that model is performing poorly.
Does it mean that there is an underlying problem ?
Summary: -0.005585 (0.234004) with: {'batch_size': 5,}
Summary: 0.849829 (0.023618) with: {'batch_size': 5, batch_size': 10, 'epochs': 100}
Summary: -0.210410 (0.131035) with: {'batch_size': 20, 'epochs': 10}
Summary: 0.294107 (0.261391) with: {'batch_size': 20, 'epochs': 50}
Summary: 0.857905 (0.006955) with: {'batch_size': 20, 'epochs': 100}
Summary: -0.381018 (0.118675) with: {'batch_size': 40, 'epochs': 10}
Summary: -0.053815 (0.321491) with: {'batch_size': 40, 'epochs': 50}
Summary: 0.582999 (0.195620) with: {'batch_size': 40, 'epochs': 100}
Summary: -0.413765 (0.114058) with: {'batch_size': 60, 'epochs': 10}
Summary: -0.018707 (0.090325) with: {'batch_size': 60, 'epochs': 50}
Summary: 0.205208 (0.157954) with: {'batch_size': 60, 'epochs': 100}
Summary: -0.423384 (0.112643) with: {'batch_size': 80, 'epochs': 10}
Summary: -0.077780 (0.101668) with: {'batch_size': 80, 'epochs': 50}
Summary: -0.000941 (0.345382) with: {'batch_size': 80, 'epochs': 100}
Summary: -0.427151 (0.112105) with: {'batch_size': 100, 'epochs': 10}
Summary: -0.177046 (0.125664) with: {'batch_size': 100, 'epochs': 50}
The Training dataset shape is (9073, 5), test: (477, 5)
Thank you
如果你对这篇内容有疑问，欢迎到本站社区发帖提问 参与讨论，获取更多帮助，或者扫码二维码加入 Web 技术交流群。
绑定邮箱获取回复消息
由于您还没有绑定你的真实邮箱，如果其他用户或者作者回复了您的评论，将不能在第一时间通知您！
发布评论
评论（1）
R2表示在回归方程创建的行中的数据点的比例。较高的R2值是可取的，因为它表明结果更好。
您需要将最小值标准化为0，最大值为1，因为R2是线路的一部分。负数是不可能的。
由于最小值为零，因此您不能在分母中使用它，因此您必须这样设置它
，因此请使用
1- [0/r1]有一个用于修复此错误
R2 indicates the proportion of data points which lie within the line created by the regression equation. A higher value of R2 is desirable as it indicates better results.
You need to standardize the minimum to 0 and the maximum to one, since r2 is a proportion of the line. A negative % is not possible.
since the minimum is zero you cant have it in the denominator, so you have to set it like this
so use
1 - [0/r1]there is a python module for fixing this error