Scribd
Upload
17 views5 pages

Spam Filtering with ML in Python

The document outlines the development of a spam filter using a machine learning algorithm, specifically Multinomial Naive Bayes, to classify emails as spam or ham. It details the preprocessing steps, model training, and inference process, along with performance evaluation metrics such as accuracy and precision. Additionally, it includes Python code for authenticating Gmail, retrieving emails, and classifying them using a pre-trained model.

Uploaded by

arasupadhmasini
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
17 views5 pages

Spam Filtering with ML in Python

The document outlines the development of a spam filter using a machine learning algorithm, specifically Multinomial Naive Bayes, to classify emails as spam or ham. It details the preprocessing steps, model training, and inference process, along with performance evaluation metrics such as accuracy and precision. Additionally, it includes Python code for authenticating Gmail, retrieving emails, and classifying them using a pre-trained model.

Uploaded by

arasupadhmasini
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd

EMAIL SPAM FILTERING USING MACHINE LEARNING IN PYTHON

Ex no : 1
Date : 20/6/25

Aim:
To develop a spam filter using a single machine learning algorithm (e.g., Naive
Bayes) trained on a labeled dataset of SMS/email messages.

Algorithm Inference:
 Model Used: Multinomial Naive Bayes (or SVM, or Random Forest)
 Input: Raw text emails (spam or ham)
 Preprocessing: Tokenization and vectorization using CountVectorizer
 Training:
o Labeled data ([Link]) is split into training/testing sets

o Text data is converted to numerical vectors

o Model is trained to identify spam keywords/patterns

 Inference:
o Given a new email message, the model transforms it to a vector
and predicts its class (spam or ham)
 Output:
o If spam: print "Deleted: <email>"

o If not spam: print "Safe: <email>"

Performance Evaluation:
 Metrics: Accuracy, Precision, Recall, F1-Score
 Observed results depend on model used but typically:
o High precision for spam class

o Fast, lightweight model


Code :
# gmail_reader.py
import os
import base64
from [Link] import Request
from [Link] import Credentials
from google_auth_oauthlib.flow import InstalledAppFlow
from [Link] import build
from classify_email import classify_email

SCOPES = ['[Link]

def authenticate_gmail():
creds = None
if [Link]('gmail_api/[Link]'):
creds = Credentials.from_authorized_user_file('gmail_api/[Link]',
SCOPES)
if not creds or not [Link]:
if creds and [Link] and creds.refresh_token:
[Link](Request())
else:
flow =
InstalledAppFlow.from_client_secrets_file('gmail_api/[Link]', SCOPES)
creds = flow.run_local_server(port=0)
with open('gmail_api/[Link]', 'w') as token:
[Link](creds.to_json())
return build('gmail', 'v1', credentials=creds)

def get_latest_emails(service, max_results=5):


results = [Link]().messages().list(userId='me',
maxResults=max_results).execute()
messages = [Link]('messages', [])
for msg in messages:
msg_data = [Link]().messages().get(userId='me',
id=msg['id']).execute()
payload = msg_data.get('payload', {})
parts = [Link]('parts', [])
data = ""

if parts:
data = parts[0]['body'].get('data', "")
else:
data = [Link]('body', {}).get('data', "")

if data:
text = base64.urlsafe_b64decode(data).decode('utf-8', errors='ignore')
print("\n--- New Email ---")
classify_email(text)
# classify_email.py

import pickle

# Load trained model and vectorizer


with open("models/spam_classifier.pkl", "rb") as f:
model = [Link](f)

with open("models/[Link]", "rb") as f:


vectorizer = [Link](f)

def classify_email(message):
vector = [Link]([message])
prediction = [Link](vector)[0]
if prediction == 1:
print(f" Deleted: {message}")
else:
print(f"✅ Safe: {message}")

# Test it directly
if __name__ == "__main__":
# Sample message to classify
test_message = "You’ve won a lottery! Click here."
classify_email(test_message)

OUTPUT :

You might also like