diff --git a/.gitignore b/.gitignore
index be20546afc6e7213216d7be700119cef0ee9d32b..2758a62a5b8721d31f9cd1dbec11c3243f475044 100644
--- a/.gitignore
+++ b/.gitignore
@@ -1,4 +1,6 @@
models/*
+results/*
+experiments/*
*/.kile/*
*.kilepr
diff --git a/code/bolsonaro/data/dataset_loader.py b/code/bolsonaro/data/dataset_loader.py
index d188cc8b18dcdf42a9d6343f03b7724acc85454a..ec1f321f70115542a2164c474193a246faa5639d 100644
--- a/code/bolsonaro/data/dataset_loader.py
+++ b/code/bolsonaro/data/dataset_loader.py
@@ -45,7 +45,7 @@ class DatasetLoader(object):
elif name == '20newsgroups_vectorized':
dataset_loading_func = fetch_20newsgroups_vectorized
task = Task.CLASSIFICATION
- elif name == 'lfw_people':
+ elif name == 'lfw_people': # needs PIL (image dataset)
dataset_loading_func = fetch_lfw_people
task = Task.CLASSIFICATION
elif name == 'lfw_pairs':
diff --git a/code/bolsonaro/models/model_parameters.py b/code/bolsonaro/models/model_parameters.py
index 768d207a323c8a7e33ed2b9b295c03cba27ce18b..60ce9ccbe10ea17cdeeaad8ade769ceaa492943a 100644
--- a/code/bolsonaro/models/model_parameters.py
+++ b/code/bolsonaro/models/model_parameters.py
@@ -5,16 +5,13 @@ import os
class ModelParameters(object):
- def __init__(self, forest_size, extracted_forest_size, normalize_D, subsets_used, seed=None):
- self._forest_size = forest_size
+ def __init__(self, extracted_forest_size, normalize_D, subsets_used, hyperparameters, seed=None):
self._extracted_forest_size = extracted_forest_size
self._normalize_D = normalize_D
self._subsets_used = subsets_used
self._seed = seed
+ self._hyperparameters = hyperparameters
- @property
- def forest_size(self):
- return self._forest_size
@property
def extracted_forest_size(self):
@@ -32,6 +29,10 @@ class ModelParameters(object):
def seed(self):
return self._seed
+ @property
+ def hyperparameters(self):
+ return self._hyperparameters
+
def save(self, directory_path, experiment_id):
save_obj_to_json(directory_path + os.sep + 'model_parameters_{}.json'.format(experiment_id),
self.__dict__)
diff --git a/code/bolsonaro/models/omp_forest_regressor.py b/code/bolsonaro/models/omp_forest_regressor.py
index 50754246abc13e0282e4cbd6aa1e917a0e3544ed..53bd76f718eea9cced6d527f67dd5ce4314e6327 100644
--- a/code/bolsonaro/models/omp_forest_regressor.py
+++ b/code/bolsonaro/models/omp_forest_regressor.py
@@ -12,8 +12,8 @@ class OmpForestRegressor(BaseEstimator):
DEFAULT_SCORE_METRIC = 'mse'
def __init__(self, models_parameters):
- self._regressor = RandomForestRegressor(n_estimators=models_parameters.forest_size,
- random_state=models_parameters.seed)
+ self._regressor = RandomForestRegressor(**models_parameters.hyperparameters,
+ random_state=models_parameters.seed)
self._models_parameters = models_parameters
self._logger = LoggerFactory.create(LOG_PATH, __name__)
@@ -82,7 +82,7 @@ class OmpForestRegressor(BaseEstimator):
self._regressor.fit(X, y)
forest = self._regressor.estimators_
return forest
-
+
def _extract_subforest(self, X, y):
"""
Given an already estimated regressor: apply OMP to get the weight of each tree.
diff --git a/code/compute_hyperparameters.py b/code/compute_hyperparameters.py
index 40e4b1d052ee026b8fe8cce8fe784a364666edbe..2c708fc5c9af04e91825a949eb57ac373aad2503 100644
--- a/code/compute_hyperparameters.py
+++ b/code/compute_hyperparameters.py
@@ -35,7 +35,7 @@ if __name__ == "__main__":
load_dotenv(find_dotenv('.env.example'))
DEFAULT_CV = 3
- DEFAULT_N_ITER = 30
+ DEFAULT_N_ITER = 50
DICT_PARAM_SPACE = {'n_estimators': Integer(10, 1000),
'min_samples_leaf': Integer(1, 1000),
'max_depth': Integer(1, 20),
@@ -56,7 +56,7 @@ if __name__ == "__main__":
begin_random_seed_range = 1
end_random_seed_range = 2000
- if args.seed is not None:
+ if args.seed is None:
random_seed = random.randint(begin_random_seed_range, end_random_seed_range)
else:
random_seed = args.seed
diff --git a/code/train.py b/code/train.py
index 546da8bfe186bf7e3f088d9ec1b74c0491f05e3a..468bfd3ab84ddecb974150747331dc0d16aaa17f 100644
--- a/code/train.py
+++ b/code/train.py
@@ -9,6 +9,7 @@ from bolsonaro.error_handling.logger_factory import LoggerFactory
from dotenv import find_dotenv, load_dotenv
import argparse
+import json
import pathlib
import random
import os
@@ -22,7 +23,6 @@ if __name__ == "__main__":
DEFAULT_DATASET_NAME = 'boston'
DEFAULT_NORMALIZE_D = False
DEFAULT_DATASET_NORMALIZER = None
- DEFAULT_FOREST_SIZE = 100
DEFAULT_EXTRACTED_FOREST_SIZE = 10
# the models will be stored in a directory structure like: models/{experiment_id}/seeds/{seed_nb}/extracted_forest_size/{nb_extracted_trees}
DEFAULT_MODELS_DIR = os.environ["project_dir"] + os.sep + 'models'
@@ -39,7 +39,7 @@ if __name__ == "__main__":
parser.add_argument('--dataset_name', nargs='?', type=str, default=DEFAULT_DATASET_NAME, help='Specify the dataset. Regression: boston, diabetes, linnerud, california_housing. Classification: iris, digits, wine, breast_cancer, olivetti_faces, 20newsgroups, 20newsgroups_vectorized, lfw_people, lfw_pairs, covtype, rcv1, kddcup99.')
parser.add_argument('--normalize_D', action='store_true', default=DEFAULT_NORMALIZE_D, help='Specify if we want to normalize the prediction of the forest by doing the L2 division of the pred vectors.')
parser.add_argument('--dataset_normalizer', nargs='?', type=str, default=DEFAULT_DATASET_NORMALIZER, help='Specify which dataset normalizer use (either standard, minmax, robust or normalizer).')
- parser.add_argument('--forest_size', nargs='?', type=int, default=DEFAULT_FOREST_SIZE, help='The number of trees of the random forest.')
+ parser.add_argument('--forest_size', nargs='?', type=int, default=None, help='The number of trees of the random forest.')
parser.add_argument('--extracted_forest_size', nargs='+', type=int, default=DEFAULT_EXTRACTED_FOREST_SIZE, help='The number of trees selected by OMP.')
parser.add_argument('--models_dir', nargs='?', type=str, default=DEFAULT_MODELS_DIR, help='The output directory of the trained models.')
parser.add_argument('--dev_size', nargs='?', type=float, default=DEFAULT_DEV_SIZE, help='Dev subset ratio.')
@@ -62,8 +62,19 @@ if __name__ == "__main__":
logger.warning('seeds and random_seed_number parameters are both specified. Seeds will be used.')
seeds = args.seeds if args.seeds is not None \
- else [random.randint(begin_random_seed_range, end_random_seed_range) \
- for i in range(args.random_seed_number)]
+ else [random.randint(begin_random_seed_range, end_random_seed_range)
+ for i in range(args.random_seed_number)]
+
+ path_hyperparameter = os.path.join('experiments', args.dataset_name, 'stage1', 'params.json')
+ if os.path.exists(path_hyperparameter):
+ with open(path_hyperparameter, 'r+') as file_hyperparameter:
+ hyperparameters = json.load(file_hyperparameter)['best_parameters']
+
+ else:
+ hyperparameters = {}
+
+ if args.forest_size is not None:
+ hyperparameters['n_estimators'] = args.forest_size
experiment_id = resolve_experiment_id(args.models_dir)
experiment_id_str = str(experiment_id)
@@ -98,11 +109,11 @@ if __name__ == "__main__":
pathlib.Path(sub_models_dir).mkdir(parents=True, exist_ok=True)
model_parameters = ModelParameters(
- forest_size=args.forest_size,
extracted_forest_size=extracted_forest_size,
normalize_D=args.normalize_D,
subsets_used=args.subsets_used,
- seed=seed
+ seed=seed,
+ hyperparameters=hyperparameters
)
model_parameters.save(sub_models_dir, experiment_id)