Spam Mail Detection

This repository contains my implementation of the Spam Mail Detection Project, completed as part of my AI & ML Internship at CODEXINTERN.
The goal is to classify email messages as Spam or Ham using Natural Language Processing (NLP) and Machine Learning techniques.

---

Exploratory Data Analysis (EDA)

Performed detailed EDA including:

• Dataset Overview: Used .head() to understand the dataset.
• Class Distribution: Analyzed the count of ham vs spam messages.
• Text Preprocessing:
  • Lowercasing text
  • Tokenization
  • Stopword removal
• Key Insights:
  • Spam messages often contain distinctive keywords and patterns.
  • Ham messages tend to be longer and more conversational.

---

Models Implemented

Two supervised learning models were trained, evaluated, and compared using Bag of Words (BoW) vectorization:

Model	Accuracy
Naive Bayes	~97%
Logistic Regression	~98%

Accuracy may vary slightly depending on preprocessing and dataset splits.

---

Evaluation

Each model was evaluated using:

• Accuracy
• Confusion Matrix (heatmap with Seaborn)
• Classification Report (Precision, Recall, F1-score)

Visual evaluations allow deeper insight into model performance and highlight strengths and weaknesses in spam detection.

---

Installation & Usage

1. Clone the repository and set up the environment

git clone https://github.com/varma1221/Spam-Mail-Detection.git
cd Spam-Mail-Detection

# Create and activate virtual environment
python3 -m venv venv
source venv/bin/activate   # On Windows use: venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt