def execute_feature_selection_a_run(run, methods):
X_train, y_train = run[0], run[2]
arr = []
for method in methods:
selector = method.split('_')[0]
n_features = int(method.split('_')[1])
arr.append(
fselect.run_feature_selection(selector, X_train, y_train,
n_features))
return arr
def normal_run(data,
n_seed=2,
splits=5,
methods=['infogain_10'],
estimators=['rdforest']):
X = np.array(data.drop('HPYLORI', axis=1))
y = np.array(data.HPYLORI)
if not os.path.exists('data'):
os.makedirs('data')
if not os.path.exists('results'):
os.makedirs('results')
features = np.zeros(X.shape[1])
outer_names = create_empty_dic(estimators, methods)
for seed in range(n_seed):
skf = StratifiedKFold(n_splits=splits, random_state=seed, shuffle=True)
split_num = 0
names = create_empty_dic(estimators, methods)
for train, test in skf.split(X, y):
split_num += 1
X_train, X_test = X[train], X[test]
y_train, y_test = y[train], y[test]
X_train, X_test = dp.impute(X_train), dp.impute(X_test)
for s in methods:
selector = s.split('_')[0]
n_features = int(s.split('_')[1])
selection = fselect.run_feature_selection(
selector, X_train, y_train, n_features)
for i in selection:
features[i] += 1
X_train, X_test = X_train[:, selection], X_test[:, selection]
for estimator in estimators:
result = e.classify(estimator, X_train, X_test, y_train,
y_test)
filename = data_path + estimator + s + '_' + str(
split_num) + '_' + str(seed)
names[estimator][s].append(filename)
result.to_csv(filename + '.csv')
create_interim_csv(names, outer_names, seed, splits)
delete_interim_csv(names)
print(outer_names)
file_list = create_final_csv(outer_names, n_seed)
return file_list
def execute_feature_selection_a_run(run, methods):
# '''
# executing feature selections of a run for the methods
# Args:
# run(list): of X_train, X_test, y_train, y_test
# methods (list): of feature selection methods
# Returns:
# List: A list of the same size as methods where each item is list of features selected for that method for that run
# '''
X_train, y_train = run[0], run[2]
arr =[]
for method in methods:
selector = method.split('_')[0]
n_features = int(method.split('_')[1])
arr.append(fselect.run_feature_selection(selector, X_train, y_train, n_features))
return arr