🩺 Cancer Risk Prediction using Machine Learning

An end-to-end Machine Learning pipeline for predicting cancer risk using patient health and lifestyle features.

---

📌 Overview

This project focuses on building and comparing multiple machine learning models for cancer risk prediction. Instead of relying on a single algorithm, different classification models, data balancing techniques, and hyperparameter tuning methods were evaluated to identify the best-performing pipeline.

---

🚀 Project Workflow

Dataset
   ↓
Exploratory Data Analysis (EDA)
   ↓
Data Cleaning
   ↓
Feature Scaling
   ↓
Model Comparison (12 Models)
   ↓
Class Imbalance Handling
   ↓
Hyperparameter Tuning
   ↓
Final Model Selection

---

📊 Models Compared

• Logistic Regression
• Decision Tree
• Random Forest
• Naive Bayes
• Support Vector Machine
• K-Nearest Neighbors
• Gradient Boosting
• AdaBoost
• XGBoost
• Bagging
• Linear Discriminant Analysis
• Quadratic Discriminant Analysis

---

⚖️ Handling Class Imbalance

The following techniques were compared:

• SMOTE
• RandomOverSampler
• ADASYN

RandomOverSampler achieved the best overall performance on this dataset.

---

🏆 Best Pipeline

RandomOverSampler
        ↓
GridSearchCV
        ↓
AdaBoost

---

📈 Final Results

Metric	Score
Accuracy	96.29%
Precision	0.96
Recall	0.97
F1 Score	0.96

---

🛠️ Technologies Used

• Python
• Pandas
• NumPy
• Matplotlib
• Seaborn
• Scikit-learn
• Imbalanced-learn
• XGBoost

---

📂 Project Structure

Cancer-Risk-Prediction-ML
│
├── dataset
├── notebook
├── images
├── README.md
├── requirements.txt
└── LICENSE

---

📌 Future Improvements

• SHAP Explainability
• LIME Explainability
• Streamlit Deployment
• Feature Importance Dashboard

---

⭐ If you found this project useful, consider giving it a star.