nguyenngocbaocmt02 / fpd

[UAI 2023] Official implementation of Efficient Failure Pattern Identification of Predictive Algorithms

[Efficient Failure Pattern Identification of Predictive Algorithms]

Python-version : 3.8.10

Generate datasets.

• run command: $ python main_src/datasets_creation_main.py <path to datasets config (.yaml)> <path to datasets save folder>
• Config of dataset in our report in dataset_creation_config.yaml, example: $ python main_src/datasets_creation_main.py main_src/datasets_creation_config.yaml datasets

Run all methods

$ chmod +x main.sh $ ./main.sh

Run Uniform sampling model

• run command: $ python main_src/ufs_main.py <the number of queries each round> <path to dataset folder> <path to result folder to save>
• In our work, the number of queries each round is 25, example: $ python main_src/ufs_main.py 25 datasets results

Run GP_CDPP model.

• run command: $ python main_src/gp_cdpp_main.py <the number of queries each round> <path to dataset folder> <path to result folder to save>
• In our work, the number of queries each round is 25, example: $ python main_src/gp_cdpp_main.py 25 datasets results

Run all visualizing script

$ chmod +x visualize.sh $ ./visualize.sh

Visualize the 2-d image of datasets

• run command: $ python visualize_src/draw_datasets.py <path to datasets folder> <path to save folder> *Example: $ python visualize_src/draw_datasets.py datasets/ figures/dataset_figures

Visualize the performance of models

• run command: $ python visualize_src/draw_performance.py <path to results folder> <path to save folder> *Example: $ python visualize_src/draw_performance.py results figures/performance_figures

Tables on report

*Example: $ python visualize_src/dataset_description_table.py datasets/ figures/datasets_description.csv *Example $ python visualize_src/performance_table.py results/ figures/performance_table.csv *Example $ python visualize_src/performance_type_table.py results/ figures/performance_type_table.csv