--- a
+++ b/split_data.py
@@ -0,0 +1,74 @@
+import os
+import numpy as np
+import random as rnd
+import pandas as pd
+
+import config
+from utils import read_csv_column, read_csv, store_to_csv
+
+
+def load_labels():
+    return read_csv(config.PATIENT_LABELS_CSV)
+
+
+def load_test_ids():
+    return read_csv_column(config.TEST_PATIENTS_IDS)
+
+
+def get_patient_name(patient_file):
+    return os.path.basename(patient_file).split('.')[0]
+
+
+def load_patient_ids():
+    test_ids = set(load_test_ids())
+    patient_ids = [get_patient_name(patient_id)
+                   for patient_id in os.listdir(config.ALL_IMGS)
+                   if get_patient_name(patient_id) not in test_ids]
+                   
+    return patient_ids
+
+def get_class(labels, patient_id):
+    return labels.get_value(patient_id, config.COLUMN_NAME)
+
+
+def count_patients_from_class(patient_ids, labels, clazz):
+    return len([patient for patient in patient_ids 
+                if get_class(labels, patient) == clazz])
+
+
+def split_data():
+    labels = load_labels()
+    total = len(labels)
+    print("Total labels loaded: ", total)
+    patient_ids = load_patient_ids()
+    print("Total patient ids loaded: ", len(patient_ids))
+    print("Patient with cancer are: ", count_patients_from_class(
+        patient_ids, labels, config.CANCER_CLS))
+
+    validation_size = int(0.15 * total)
+
+    validation_set = rnd.sample(patient_ids, validation_size)
+    train_set = [patient for patient in patient_ids 
+                 if patient not in validation_set]
+
+    print("Patients for training: ", len(train_set))
+    print("Patients for validation: ", len(validation_set))
+
+    print("Patients with cancer in validation set {}, no cancer {}.".format(
+        count_patients_from_class(validation_set, labels, config.CANCER_CLS),
+        count_patients_from_class(validation_set, labels, config.NO_CANCER_CLS)))
+    print("Patients with cancer in training set {}, no cancer {}.".format(
+        count_patients_from_class(train_set, labels, config.CANCER_CLS),
+        count_patients_from_class(train_set, labels, config.NO_CANCER_CLS)))
+
+    validation_labels = [get_class(labels, p) for p in validation_set]
+    store_to_csv(validation_set, validation_labels, 
+        config.VALIDATION_PATINETS_IDS)
+    
+    train_labels = [get_class(labels, p) for p in train_set]
+    store_to_csv(train_set, train_labels, 
+        config.TRAINING_PATIENTS_IDS)
+
+
+if __name__ == '__main__':
+    split_data()
\ No newline at end of file