MSc in FE Course - 6. Machine Learning in Finance - M7: Hyperparameter Tuning and Model Optimization

1. Before deploying into Prod: ensure the model is devoid of bias and generalizes well on new unseen data.

2.  We can the train, validation, and test set by:

  • Random Split
  • Stratified Split

3. Random Split -

4. Stratified Split -  divide data into multiple subsets while ensuring that the proportions of classes or categories are preserved across those subsets.

5.


References

  1. Yu, Tong, and Hong Zhu. "Hyper-Parameter Optimization: A Review of Algorithms and Applications." arXiv, 2020, https://arxiv.org/abs/2003.05689.

Subscribe to: Posts (Atom)