Evaluation and selection of ESL 7 models _ reading notes

Bias, variance and Model complexity

The predicted value and the real merit are two kinds of functions generally:

The error of inspection or generalization is the prediction error on independent prediction Sample:

There are two objectives for the evaluation and selection of models: Model Evaluation: Comparing the performance of multiple models to select the optimal model evaluation: Estimating its generalization error in the test set

Data sets are generally divided into: training set, validation set, test set bias variance decomposition

For K-neareast There is a brief form:

For linear fitting