Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Learning policy fix #246

Open
wants to merge 7 commits into
base: main
Choose a base branch
from
Open

Learning policy fix #246

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
e266264
minor
maypink Dec 7, 2023
1fe1c0d
add mabs saving
maypink Dec 7, 2023
a6f2b90
fix minor with saving
maypink Dec 9, 2023
0c4986c
reverse learnong policy
maypink Dec 13, 2023
7ec715f
minor
maypink Dec 14, 2023
551a3f6
roll back later
maypink Dec 14, 2023
66f3014
ucb to 0.8
maypink Dec 17, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
102 changes: 51 additions & 51 deletions examples/experiment_analyzer/experiment_analyzer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -12,71 +12,71 @@
""" The result of analysis can be seen without running the script in
'~/GOLEM/examples/experiment_analyzer/result_analysis.tar.gz'. """

path_to_root = os.path.join(project_root(), 'examples', 'experiment_analyzer')
path_to_root = os.path.join('Z://', 'Pinchuk')

# extract data if there is an archive
if 'data.tar.gz' in os.listdir(path_to_root):
tar = tarfile.open(os.path.join(path_to_root, 'data.tar.gz'), "r:gz")
tar.extractall()
tar.close()

path_to_experiment_data = os.path.join(path_to_root, 'data')
path_to_experiment_data = path_to_root
path_to_save = os.path.join(path_to_root, 'result_analysis')

analyzer = ExperimentAnalyzer(path_to_root=path_to_experiment_data, folders_to_ignore=['result_analysis',
'Thumbs.db'])

# to get convergence table with mean values
path_to_save_convergence = os.path.join(path_to_save, 'convergence')

convergence_mean = analyzer.analyze_convergence(history_folder='histories', is_raise=False,
path_to_save=path_to_save_convergence,
is_mean=True)

# to get convergence boxplots
convergence = analyzer.analyze_convergence(history_folder='histories', is_raise=False)
path_to_save_convergence_boxplots = os.path.join(path_to_save_convergence, 'convergence_boxplots')

metrics = list(convergence.keys())
setups = list(convergence[metrics[0]].keys())
datasets = list(convergence[metrics[0]][setups[0]].keys())
for dataset in datasets:
for metric_name in convergence.keys():
to_compare = dict()
for setup in convergence[metric_name].keys():
to_compare[setup] = [i for i in convergence[metric_name][setup][dataset]]
plt.boxplot(list(to_compare.values()), labels=list(to_compare.keys()))
plt.title(f'Convergence on {dataset}')
os.makedirs(path_to_save_convergence_boxplots, exist_ok=True)
plt.savefig(os.path.join(path_to_save_convergence_boxplots, f'convergence_{dataset}.png'))
plt.close()
#
# # to get convergence table with mean values
# path_to_save_convergence = os.path.join(path_to_save, 'convergence')
#
# convergence_mean = analyzer.analyze_convergence(history_folder='histories', is_raise=False,
# path_to_save=path_to_save_convergence,
# is_mean=True)
#
# # to get convergence boxplots
# convergence = analyzer.analyze_convergence(history_folder='histories', is_raise=False)
# path_to_save_convergence_boxplots = os.path.join(path_to_save_convergence, 'convergence_boxplots')
#
# metrics = list(convergence.keys())
# setups = list(convergence[metrics[0]].keys())
# datasets = list(convergence[metrics[0]][setups[0]].keys())
# for dataset in datasets:
# for metric_name in convergence.keys():
# to_compare = dict()
# for setup in convergence[metric_name].keys():
# to_compare[setup] = [i for i in convergence[metric_name][setup][dataset]]
# plt.boxplot(list(to_compare.values()), labels=list(to_compare.keys()))
# plt.title(f'Convergence on {dataset}')
# os.makedirs(path_to_save_convergence_boxplots, exist_ok=True)
# plt.savefig(os.path.join(path_to_save_convergence_boxplots, f'convergence_{dataset}.png'))
# plt.close()

# to get metrics table with mean values
path_to_save_metrics = os.path.join(path_to_save, 'metrics')
metric_names = ['roc_auc', 'f1']
path_to_save_metrics = os.path.join(path_to_save, 'metrics_with_UCB')
metric_names = ['roc_auc', 'f1', 'logloss']
metrics_dict_mean = analyzer.analyze_metrics(metric_names=metric_names, file_name='evaluation_results.csv',
is_raise=False, path_to_save=path_to_save_metrics,
is_mean=True)

# to get metrics boxplots
metrics_dict = analyzer.analyze_metrics(metric_names=metric_names, file_name='evaluation_results.csv',
is_raise=False)
path_to_save_metrics_boxplots = os.path.join(path_to_save_metrics, 'metrics_boxplot')

for metric in metric_names:
for dataset in metrics_dict[metric][list(metrics_dict[metric].keys())[0]].keys():
to_compare = dict()
for setup in metrics_dict[metric].keys():
to_compare[setup] = [-1 * i for i in metrics_dict[metric][setup][dataset]]
plt.boxplot(list(to_compare.values()), labels=list(to_compare.keys()))
plt.title(f'{metric} on {dataset}')
cur_path_to_save = os.path.join(path_to_save_metrics_boxplots, metric)
os.makedirs(cur_path_to_save, exist_ok=True)
plt.savefig(os.path.join(cur_path_to_save, f'{metric}_{dataset}.png'))
plt.close()

# to get stat test results table
path_to_save_stat = os.path.join(path_to_save, 'statistic')
stat_dict = analyzer.analyze_statistical_significance(data_to_analyze=metrics_dict['roc_auc'],
stat_tests=[mannwhitneyu, kruskal, ttest_ind],
path_to_save=path_to_save_stat)
# # to get metrics boxplots
# metrics_dict = analyzer.analyze_metrics(metric_names=metric_names, file_name='evaluation_results.csv',
# is_raise=False)
# path_to_save_metrics_boxplots = os.path.join(path_to_save_metrics, 'metrics_boxplot')
#
# for metric in metric_names:
# for dataset in metrics_dict[metric][list(metrics_dict[metric].keys())[0]].keys():
# to_compare = dict()
# for setup in metrics_dict[metric].keys():
# to_compare[setup] = [-1 * i for i in metrics_dict[metric][setup][dataset]]
# plt.boxplot(list(to_compare.values()), labels=list(to_compare.keys()))
# plt.title(f'{metric} on {dataset}')
# cur_path_to_save = os.path.join(path_to_save_metrics_boxplots, metric)
# os.makedirs(cur_path_to_save, exist_ok=True)
# plt.savefig(os.path.join(cur_path_to_save, f'{metric}_{dataset}.png'))
# plt.close()
#
# # to get stat test results table
# path_to_save_stat = os.path.join(path_to_save, 'statistic')
# stat_dict = analyzer.analyze_statistical_significance(data_to_analyze=metrics_dict['roc_auc'],
# stat_tests=[mannwhitneyu, kruskal, ttest_ind],
# path_to_save=path_to_save_stat)
2 changes: 2 additions & 0 deletions experiments/experiment_analyzer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -409,6 +409,8 @@ def _get_path_to_launch(self) -> Tuple[str, str, str]:
for setup in os.listdir(self.path_to_root):
if setup in self._folders_to_ignore:
continue
if setup != 'FEDOT_MAB_UCB' and setup != 'FEDOT_MAB' and setup != 'FEDOT_Classic':
continue
path_to_setup = os.path.join(self.path_to_root, setup)
for dataset in os.listdir(path_to_setup):
if dataset in self._folders_to_ignore:
Expand Down
4 changes: 3 additions & 1 deletion golem/core/optimisers/adaptive/mab_agents/contextual_mab_agent.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -34,7 +34,7 @@ def __init__(self, actions: Sequence[ActType],
super().__init__(actions=actions, n_jobs=n_jobs, enable_logging=enable_logging,
decaying_factor=decaying_factor, is_initial_fit=False)
self._agent = MAB(arms=self._indices,
learning_policy=LearningPolicy.UCB1(alpha=1.25),
learning_policy=LearningPolicy.UCB1(alpha=0.8),
neighborhood_policy=NeighborhoodPolicy.Clusters(),
n_jobs=n_jobs)
self._context_agent = context_agent_type if isinstance(context_agent_type, Callable) else \
Expand Down Expand Up @@ -81,6 +81,8 @@ def partial_fit(self, experience: ExperienceBuffer):
obs, arms, processed_rewards = self._get_experience(experience)
contexts = self.get_context(obs=obs)
self._agent.partial_fit(decisions=arms, rewards=processed_rewards, contexts=contexts)
if self._path_to_save:
self.save()

def _get_experience(self, experience: ExperienceBuffer):
""" Get experience from ExperienceBuffer, process rewards and log. """
Expand Down
2 changes: 1 addition & 1 deletion golem/core/optimisers/adaptive/mab_agents/mab_agent.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -82,7 +82,7 @@ def save(self, path_to_save: Optional[str] = None):
if not path_to_save.endswith('.pkl'):
os.makedirs(path_to_save, exist_ok=True)
mabs_num = [int(name.split('_')[0]) for name in os.listdir(path_to_save)
if re.fullmatch(r'\d_mab.pkl', name)]
if re.fullmatch(r'\d*_mab.pkl', name)]
if not mabs_num:
max_saved_mab = 0
else:
Expand Down
Loading