class Experiment:
def __init__(self):
self.data_loader = DataLoader('/data')
self.trainer = Trainer()
def run(self):
train_data, train_labels = self.data_loader.load_set('train')
test_data, test_labels = self.data_loader.load_set('test')
train_data = self.data_loader.clean_data(train_data)
test_data = self.data_loader.clean_data(test_data)
self.trainer.train(train_data, train_labels)
train_predictions = self.trainer.predict(train_data)
test_predictions = self.trainer.predict(test_data)
print('Train accuracy: ',
self.compute_score(train_labels, train_predictions))
print('Test accuracy: ',
self.compute_score(test_labels, test_predictions))
def compute_score(self, labels, predictions):
return metrics.accuracy_score(labels, predictions)
method and compares them
'''
#%%
# Read the data:
data = DataLoader(dataset_name='trade_selection', extension='csv')
split = int(data.df.shape[0] * 0.8)
# Compute a threshold to create the label:
threshold_ = analysis_threshold(data.df, split)
# Some plots of the variable Result:
plot_result(data.df)
plot_hist(data.df)
# Clean the data:
datas = data.clean_data(threshold=threshold_)
old_df = datas.copy()
datas = datas.drop(['Result'], axis=1)
# Split the train_test and evaluation set:
train_evaluation_set = datas.iloc[:split, :]
test_set = datas.iloc[split:, :]
# Get the block variables:
list_var = data.get_list_var_block()
#%% OCA method
oca_method = OCAMethod(train_evaluation_set,
n_min=10,
verbose=True,
reload_feature_importance=False,
@author: david.saltiel
"""
from data_loader import DataLoader
from OCA import OCAMethod
from RFE import RFEMethod
from BCA import BCAMethod
from graphics import create_graphic_for_method, create_figure2, create_figure3
''' This is the main function
It calls all the 3 features selection
method and compares them
'''
#%% reads the data
data = DataLoader(dataset_name = 'trade_selection_A229', extension = 'csv')
data.clean_data()
#%% OCA method
oca_method = OCAMethod(data.df, n_min=10, verbose = True,reload_feature_importance = False)
oca_method.select_features()
#%% RFE method
rfe_method = RFEMethod(data.df, verbose = True)
rfe_method.select_features()
rfe_dictionary = rfe_method.save_all_score()
#%% BCA method
bca_method = BCAMethod(data.df, verbose = True)
bca_method.select_features()
