A Comparative Study of Hyperparameter Tuning Methods

The study emphasizes the challenge of finding the optimal trade-off between bias and variance, especially as hyperparameter optimization increases in complexity. Through empirical analysis, three hyperparameter tuning algorithms Tree-structured Parzen Estimator (TPE), Genetic Search, and Random Search are evaluated across regression and classification tasks. The results show that nonlinear models, with properly tuned hyperparameters, significantly outperform linear models. Interestingly, Random Search excelled in regression tasks, while TPE was more effective for classification tasks. This suggests that there is no one-size-fits-all solution, as different algorithms perform better depending on the task and model type. The findings underscore the importance of selecting the appropriate tuning method and highlight the computational challenges involved in optimizing machine learning models, particularly as search spaces expand.
Comment: This chapter has been accepted in the edited volume titles "Data Science in Theory and Practice", editor J Sen & S Roy Choudhury. The volume is expected to be published in October 2024 by Cambridge Scholars Publishing, New Castle upon Tyne, UK. This chapter is 34 pages long and it contains 11 tables and 8 images