Breast Cancer Prediction Web Application

🚀 Live App: breast-cancer-prediction-app-by-rudra.streamlit.app

---

---

📌 Overview

This is a Streamlit-based web application that predicts whether a breast tumor is benign or malignant using a trained logistic regression model. Built using the Breast Cancer Wisconsin Diagnostic Dataset, this project demonstrates the complete data science lifecycle — from problem definition to model deployment.

🎯 Objective: Build a reliable, interactive, and modular diagnostic tool that aids in early breast cancer detection using medical imaging features.

---

⭐ Project Highlights

• ✅ Demonstrates end-to-end ML workflow
• ✅ Real-time prediction app with deployed UI
• ✅ Modular, readable, and scalable Python code
• ✅ Excellent use case for health tech/AI in diagnostics

---

🧠 Key Features

• Predict tumor type (Benign or Malignant)
• Real-time probability scores
• User-friendly UI with input sliders
• Modular codebase for easy scalability
• Live deployment on Streamlit Cloud

---

🗂️ Project Architecture

├── assets/                # Visual assets (e.g., logos, screenshots)
├── data/                 # Dataset and derived files
├── model/                # Trained model (Pickle file)
├── notebooks/            # Exploratory and preprocessing notebooks
│   ├── p1-understand-the-data.ipynb
│   ├── p2-eda.ipynb
│   └── p3-outliers.ipynb
├── utils/                # Custom modules for charts, sidebar, data
│   ├── charts.py
│   ├── data_model.py
│   ├── sidebar.py
├── streamlit_app.py      # Main app entry point
├── requirements.txt      # Python dependencies
└── README.md             # Project documentation

---

📊 Features Used for Prediction

The model uses 30 key measurements from cell nuclei obtained via digitized images, including:

• Mean: Radius, Texture, Perimeter, Area, Smoothness
• Standard Error: Radius SE, Perimeter SE, Concavity SE, etc.
• Worst (largest): Texture worst, Area worst, Symmetry worst, etc.

Each of these is captured using an intuitive sidebar interface in the app.

---

🧪 Model Details

• Algorithm: Logistic Regression
• Library: scikit-learn
• Training Dataset: UCI Breast Cancer Diagnostic Data
• Model File: Saved as model.pkl for production use

Note: The pickle file is version-sensitive. Please ensure compatible scikit-learn versions (1.6.1 recommended) for loading the model.

---

📌 Step-by-Step Workflow

This project was developed following an end-to-end data science pipeline:

1. Problem Definition
2. Data Collection & Cleaning
3. Exploratory Data Analysis (EDA)
4. Outlier Handling & Feature Engineering
5. Model Building & Evaluation
6. Prediction Pipeline Creation
7. Web App Development using Streamlit
8. Deployment on Streamlit Cloud

---

🧰 Tech Stack

Tool/Library	Usage
Python	Core programming language
Pandas / NumPy	Data manipulation
Matplotlib / Seaborn	Data visualization
Plotly	Interactive Data visualization
Scikit-learn	Model building & evaluation
Streamlit	Web interface & deployment
Pickle	Model serialization

---

📷 App Interface

Example inputs include cell features like radius, area, smoothness, etc.
The prediction result is shown instantly with confidence levels.

cancer-prediction.mp4

---

🧪 Run the App Locally

• Clone the repository

  git clone https://github.com/Rudra-G-23/breast-cancer-prediction-app.git
  cd breast-cancer-prediction-app

• Install dependencies

  pip install -r requirements.txt

• Launch the app

  streamlit run streamlit_app.py

---

📚 References

• 📘 Streamlit Deployment Guide
• 📊 Kaggle Dataset
• 🔬 Feature Engineering Notebook
• 📈 Modeling Example
• 🎥 Inspiration

---

🙋‍♂️ Author

Rudra Prasad Bhuyan
📧 rudraprasadbhuyan000@gmail.com
🔗 GitHub | LinkedIn | Kaggle

---