Error
Estimator(估算器）
From training data we find f* , f* is an estimator of
Bias and Variance of Estimator
N比较大时，比较集中；N比较小时，比较分散。
Varience
Simple model is less influenced by the sampled data
Bias
简单的model有比较大的Bias，复杂的model有比较小的Bias
关键问题：你现在的model是Bias大还是Variance大？
regularization 时可能会伤害到Bias
Model Selection
- There is usually a trade-off between bias and variance.
- Select a model that balances two kinds of error to minimize total error
- what you should not do :
- 使用 交叉验证 来处理上述这种情况
-
- 注意不要 根据 Testing Set 去调参，因为现在的Testing Set 不一定就是真正的 Testing Set，你根据这个Testing Set 调参会把这个Testing Set 的偏差考虑进去，影响到model。