elasticnet_train_sklearn.seq

import sys
import python

pydef get_all_training_files_in_folder(directory) -> list[tuple[str, str]]:
    import os
    files = []
    for item in os.listdir(directory):
        item_path = os.path.join(directory, item)
        if os.path.isfile(item_path):
            if item.endswith(".csv"):
                files.append((item_path, item_path[:-4] + ".labels"))
    return files

pydef training(directory, train_files, l1, l2):
    import os
    import numpy as np
    from sklearn.linear_model import ElasticNet
    from sklearn.preprocessing import StandardScaler
    y_list = []
    X_list = []
    for csv_f, lbl_f in train_files:
        with open(lbl_f) as lbl_file:
            for line in lbl_file:
                age, Fage = line.rstrip().split(',')
                y_list.append(Fage)
        with open(csv_f) as csv_file:
            X = np.genfromtxt(csv_file, dtype=float, delimiter=',')
            X_list.append(X)
    X_train = np.concatenate(X_list)
    scaler = StandardScaler()
    X_train = scaler.fit_transform(X_train)
    y_train = np.array(y_list, dtype=np.float)
    regr = ElasticNet(random_state=0, alpha=l2, l1_ratio=l1, normalize=False,
                        fit_intercept=True, max_iter=10000)
    regr.fit(X_train, y_train)
    np.save(os.path.join(directory, "enet_sk_betas.npy"), regr.coef_)
    np.save(os.path.join(directory, "enet_sk_intercept.npy"), regr.intercept_)

input_folder = sys.argv[1]
save_folder = sys.argv[2]
#L1 regularization term alpha
l1 = 0.02255706
#L2 regularization term lambda
l2 = 0.5
train_files = get_all_training_files_in_folder(input_folder)
training(save_folder, train_files, l1, l2)