PRZI-Bayesian-Optimisation

The BSE.py is refered from Bristol Stock Exchange. We add our proposed PRBO algorithm to it.

Installation

git clone https://github.com/HarmoniaLeo/PRZI-Bayesian-Optimisation
cd PRZI-Bayesian-Optimisation
conda create --name PRZI-Bayesian-Optimisation --file requirements.txt
conda active PRZI-Bayesian-Optimisation

Hyper-parameter Selections

To run the Hyperparameters Selection for PRSH, firstly run the command:

python PRSH_hyper-parameters-experiments.py

The result can be found in results/PRSH_results.csv. Then use PRSH_hyper-parameters-statistics.ipynb to perform hypothesis tests and select the possibly optimized hyper-parameter combinations.

To run the Hyperparameters Selection for PRBO, firstly:

python PRBO_hyper-parameters-experiments.py

The result can be found in results/PRBO_results.csv. Then use PRBO_hyper-parameters-statistics.ipynb to perform hypothesis tests and select the possibly optimized hyper-parameter combinations.

Intermediate result files generated by BSE will be saved in results_buffer.

Comparison between PRBO and PRSH

To run the comparison experiments, firstly run the command:

python PRBOvsPRSH_experiments.py

The result can be found in results/PRBOvsPRSH_results.csv. Then use PRBOvsPRSH_statistics.ipynb to perform hypothesis tests and view the comparison results.

Intermediate result files generated by BSE will be saved in results_buffer.