Spam SMS Classification Project 📱🕵️‍♂️

Overview

This project aims to classify SMS messages as spam or non-spam using natural language processing (NLP) and machine learning techniques. We use the TfidfVectorizer for text preprocessing and various machine learning algorithms to identify and categorize SMS content effectively.

Project link Spam_SMS_Classification

Table of Contents

• Getting Started 🚀
• Prerequisites 📋
• Data 📊
• Text Preprocessing 📝
• Model Training 🏋️‍♂️
• Evaluation 📈
• Class Imbalance & SMOTE ⚖️
• Optimization 🛠️
  • Grid Search 🔍
  • Random Search 🎲
• Real World Examples 🌍
• Results 🏆
• Contributing 🤝
• License 📄
• Acknowledgments 🙏

Getting Started 🚀

Follow these instructions to set up and run the project locally for development and testing purposes.

Prerequisites 📋

Ensure that you have Python installed along with the required libraries. Download Python from python.org and install libraries using pip.

Data 📊

The dataset for this project is located in the data directory. The primary file is sms_spam.csv, which contains SMS messages labeled as spam or ham (non-spam). The dataset is sourced from Kaggle.

Text Preprocessing 📝

We use the TfidfVectorizer to preprocess SMS messages, transforming raw text into numerical features that machine learning algorithms can use.

1. TF-IDF Vectorization:

   • Term Frequency-Inverse Document Frequency (TF-IDF) is a technique that converts text into numerical features based on the importance of each word in the document relative to the entire corpus.
   • Term Frequency (TF) measures how often a word appears in a document.
   • Inverse Document Frequency (IDF) adjusts the term frequency by how often the word appears across all documents. It reduces the weight of common words and increases the weight of rare words.
   • TF-IDF Score is the product of TF and IDF, representing the importance of a word in a document.

   Example:

   • For the message "Win a $100 gift card", TF-IDF will assign higher scores to less common words like "gift" and "card" compared to common words like "win" and "a".

Model Training 🏋️‍♂️

The model training process is detailed in the model_train.py script. We utilize decision tree and random forest algorithms to train models on the preprocessed SMS data.

Decision Tree:

• A popular classification model that splits the data into smaller regions based on feature values, making predictions based on these splits.
• Decision trees are easy to interpret and visualize, making them useful for understanding the model's decision-making process.

Random Forest:

• An ensemble learning method that combines multiple decision trees to improve accuracy and robustness.
• Random forests aggregate predictions from multiple trees to provide a final classification decision.

Evaluation 📈

The performance of the models is evaluated using metrics such as accuracy, precision, recall, and F1-score. Detailed results are available in the model_evaluation.py script.

Class Imbalance & SMOTE ⚖️

Spam SMS classification often faces class imbalance, where one class (e.g., spam) may be underrepresented compared to the other (e.g., ham). To address this:

1. Class Imbalance:

   • Class imbalance can lead to biased models that perform poorly on the minority class.
   • Evaluating class distribution helps in understanding the extent of imbalance.

2. SMOTE (Synthetic Minority Over-sampling Technique):

   • SMOTE generates synthetic samples to balance the class distribution.
   • By creating new instances of the minority class, SMOTE helps the model learn better and improve classification performance.

Visualization:

• Class Distribution Before SMOTE: Visualize the imbalance in the dataset.
• Class Distribution After SMOTE: Show the balanced distribution post-SMOTE application.

Optimization 🛠️

We optimize our models using hyperparameter tuning techniques to enhance performance.

Grid Search 🔍

• Grid Search exhaustively searches through a specified range of hyperparameters to find the best combination.
• It evaluates all possible combinations within the defined range and selects the optimal set based on performance metrics.

Random Search 🎲

• Random Search samples random combinations of hyperparameters from a defined range.
• It is often faster than Grid Search and can efficiently find good hyperparameters without exhaustive searching.

Visualization:

• Grid Search Optimization Results: Display the best parameters found through Grid Search.
• Random Search Optimization Results: Show the best parameters found through Random Search.

Real World Examples 🌍

Here we showcase examples of spam and non-spam SMS messages, demonstrating the model's performance in real-world scenarios.

1. Example 1: "Congratulations! You've won a $1000 gift card. Call now to claim your prize!"
2. Example 2: "Hey, how are you? Let's catch up this weekend."
3. Example 3: "Your account has been compromised. Click this link to secure it immediately."
4. Example 4: "Are you available for a quick meeting tomorrow?"
5. Example 5: "Limited time offer! Buy one get one free on all items."

Visualization:

• Classification Report: Display the classification report, showing metrics such as precision, recall, and F1-score for the real-world examples.

Results 🏆

The trained models are saved in the models directory. You can use these models to classify new SMS messages.

Contributing 🤝

We welcome contributions to this project! Please open an issue or create a pull request if you'd like to contribute.

License 📄

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgments 🙏

Kaggle