Train_Test_fold_loop

Train_Test_fold_loop.py

df = pd.read_excel(r"C:\Rhythm\test_ds_loc.xlsx")
unique_subjects = df['p_id'].unique()

for subject in unique_subjects:
    test_subject_df = df[df['p_id'] == subject].sort_values(by='date')
    train_df = df[df['p_id'] != subject]

    n = len(test_subject_df)
    if n < 3:
        continue

    split1 = n // 3
    split2 = 2 * n // 3

    test_fold = test_subject_df.iloc[split2:]
    pseudo_train_fold = test_subject_df.iloc[:split2]

    final_train_df = pd.concat([train_df, pseudo_train_fold], ignore_index=True)

    td = final_train_df[final_train_df['p_id'] == subject]
    print(f"Subject: {subject} | last Train Date: {td['date'].max()} | 1st Test Date: {test_fold['date'].min()}")