Begin work on email dataset loading

Author: maillouxc

part2/experiments.py

@@ -1,15 +1,32 @@
 #!/usr/bin/env python3
 
 import argparse
+import os
+import glob
 from time import time
 
 from sklearn import svm
 from sklearn.linear_model import Perceptron
 
 
 def main():
- parsedArgs = parse_command_line_arguments()
- load_dataset(parsedArgs.path)
+ parsed_args = parse_command_line_arguments()
+
+ data_path = parsed_args.path
+ training_emails, training_labels = load_dataset(data_path + "/training")
+ test_emails, test_labels = load_dataset(data_path + "/test")
+
+ print("Training Emails data:")
+ print(training_emails)
+ print()
+ print("Training Emails labels:")
+ print(training_labels)
+ print()
+
+ svm_classifier = train_svm_classifier(training_emails, training_labels)
+
+ results = svm_classifier.predict(test_emails)
+ print(results)
 
 
 def parse_command_line_arguments():
@@ -22,8 +39,31 @@ def parse_command_line_arguments():
  return parser.parse_args()
 
 
-def load_dataset(dataPath):
- print(dataPath)
+def load_dataset(data_path):
+ print("Loading dataset at " + data_path)
+
+ email_data = []
+ email_labels = []
+
+ os.chdir(data_path)
+ for file in os.listdir():
+ if (file.startswith("sp")):
+ email_labels.append("spam")
+ else:
+ email_labels.append("ham")
+
+ print("Loaded " + str(len(email_labels)) + " emails")
+
+ return email_data, email_labels
+
+
+def train_svm_classifier(emails, labels):
+ classifier = svm.SVC(
+ gamma='scale',
+ C=1,
+ verbose=True)
+ classifier.fit(emails, labels)
+ return classifier
 
 
 if __name__ == "__main__":