MS

Summary

Model Selection & Evaluation

• Evaluation Method
  • Continuous: MSE
  • Classification: Accuracy, Entropy, Precision & Recall & TPR & FPR & F1 score, ROC & AUC & cost curve

Overfitting

• Training Error/Empirical Error
• Generalization Error
• When learner learns about specific tasks too much(lower training error), it may decrease the ability on learning general tasks(higher generalization error)

Bias-Variance Trade-off

Lectures & Readings

Demonstration

• Implement Cross Validation with Python (Cross Validation.ipynb)