Spam Detection Model

An advanced spam detection model using Naive Bayes and TF-IDF vectorization for accurate spam filtering. Built with FastAPI and deployed with Uvicorn.

---

🌐 Live Demo

👉 HiveMind Spam Detector – Try it live!

---

🚀 Features

✅ Classifies messages as Spam or Not Spam
✅ High accuracy with TF-IDF + Multinomial Naive Bayes
✅ Pre-trained model stored in .pkl format for fast loading
✅ FastAPI-based REST API with minimal latency

---

🏗️ Project Structure

📂 HiveMind-Spam-Detection-ML ├── 📁 app/ # FastAPI app and routes ├── 📁 data/ # Training data (CSV) ├── 📁 models/ # Pre-trained ML models (.pkl) ├── 📄 requirements.txt # Dependencies ├── 📄 README.md # Project documentation

---

🛠️ Setup

1️⃣ Clone the repository

git https://github.com/Anshu-x/HiveMind-Spam-Detection-ML.git
cd HiveMind-Spam-Detection-ML

2️⃣ Create a virtual environment

python -m venv venv
source venv/bin/activate # For Linux/MacOS
venv\Scripts\activate # For Windows

3️⃣ Install dependencies

pip install -r requirements.txt
🚦 Run the App Start the FastAPI server using Uvicorn:

uvicorn app.main:app --host 0.0.0.0 --port 8000 The app will be available at: http://localhost:8000/docs

🔍 Making Predictions

Send a POST request to the /predict endpoint:

Example using cURL:

curl -X POST http://localhost:8000/predict
-H "Content-Type: application/json"
-d '{ "message": "You have won a free iPhone! Click the link to claim it." }' Example using Python Requests:

import requests

url = "http://localhost:8000/predict" data = {"message": "You have won a free iPhone! Click the link to claim it."} response = requests.post(url, json=data) print(response.json()) ✅ Sample Response

{ "prediction": "Spam", "confidence": 0.92 } 🏋️ Training Train the model using the dataset in the /data folder:

python app/train.py 👉 Adjust hyperparameters like alpha in train.py:

from sklearn.naive_bayes import MultinomialNB

model = MultinomialNB(alpha=0.1) 👉 Save the model using joblib:

import joblib

joblib.dump(model, 'models/spam_detector.pkl')

💻 Code Overview

✅ Spam Classification Logic

Uses TF-IDF vectorization for feature extraction Multinomial Naive Bayes for classification

from sklearn.feature_extraction.text import TfidfVectorizer from sklearn.naive_bayes import MultinomialNB import joblib

vectorizer = TfidfVectorizer() X_train = vectorizer.fit_transform(data['message']) y_train = data['label']

model = MultinomialNB(alpha=0.1) model.fit(X_train, y_train)

#Save model joblib.dump(model, 'models/spam_detector.pkl')

✅ FastAPI Endpoint

/predict → POST request Accepts message string Returns classification and confidence score

from fastapi import FastAPI from pydantic import BaseModel import joblib

app = FastAPI()

model = joblib.load('models/spam_detector.pkl') vectorizer = joblib.load('models/tfidf_vectorizer.pkl')

class Message(BaseModel): message: str

@app.post('/predict') def predict(msg: Message): input_vector = vectorizer.transform([msg.message]) prediction = model.predict(input_vector)[0] confidence = model.predict_proba(input_vector).max() return {"prediction": prediction, "confidence": confidence}

📊 Performance

Metric Value Accuracy 96.5% Precision 95.8% Recall 94.6% F1-Score 95.2%

🎯 Tech Stack

Tool/Library Purpose Python 3.10+ Core Programming Language FastAPI Web Framework Scikit-learn Machine Learning Joblib Model Serialization Pandas Data Handling Uvicorn ASGI Server

🚨 To-Do

Improve classification for multilingual input Implement streaming classification Add logging and monitoring

💡 Contributions welcome! Feel free to submit a PR or open an issue. 👊

---