Email Spam Classification using Machine Learning

📌 Overview

This project classifies emails as Spam or Not Spam using structured email campaign data. Machine learning models are trained using numerical and categorical features derived from email campaigns.

---

🎯 Problem Statement

Email spam impacts communication efficiency and user trust. The goal of this project is to predict whether an email is spam based on campaign-related attributes.

---

📊 Dataset Description

The dataset contains campaign-level features related to emails.

Input Features

• Email_Type
• Subject_Hotness_Score
• Email_Source_Type
• Customer_Location
• Email_Campaign_Type
• Total_Past_Communications
• Time_Email_sent_Category
• Word_Count
• Total_Links
• Total_Images

Target Variable

• Email_Status
  • 0 → Not Spam
  • 1 → Spam

---

🛠️ Methodology

1. Data inspection and preprocessing
2. Handling categorical and numerical features
3. Feature-target separation
4. Train-test split
5. Classification model training
6. Model evaluation

---

🤖 Models Used

• Logistic Regression
• Naive Bayes

---

📈 Evaluation Metrics

• Accuracy
• Precision
• Recall
• F1-score

---

🔍 Key Insights

• Subject hotness score impacts spam classification
• Link and image count influence email status
• Simple classifiers perform well on structured data

---

🚀 Future Enhancements

• Advanced feature engineering
• Try ensemble classifiers
• Apply explainability (SHAP)
• Deploy as a web application

---

🧠 Learnings

• Binary classification on structured datasets
• Handling categorical campaign features
• Model evaluation for imbalance scenarios