Project for the course of Statistical Methods for Machine Learning in which multiple machine learning algorithms are discussed and trained in order to fit the given dataset. The algorithms are then evaluated with expanded features and kernel methods (Polynomial and Guassian kernels).
- The hyperparameter tuning is done using Bayesian Optimization. For better understanding the optimization method used, I suggest reading this blog post. module skopt is used to implement Bayesian Optimization.
- In the implementations of the methods in this project, no machine learning modules have been used and all are written from scratch.
A full report of the project is also provided including the explanations, performance results, computational complexity and comparisons between different methods.