AlgoFest_CancerPrediction

• 

  Logistic Regression- Confusion Matrix Heatmap

• 

  Logistic Regression- ROC Curve

• 

  Decision Tree- Confusion Matrix

• 

  Decision Tree- ROC Curve

• 

  Random Forest- Confusion Matrix

• 

  Random Forest- ROC Curve

• 

  Model AUC Comparison

• 

  Feature Importance- Random Forest Model

• 

  Prediction by setting features' values

• 

  Breast Cancer Prediction Dashboard

Inspiration

This project is deeply personal. When I was 9, my aunt was diagnosed with cancer and unfortunately could not survive. She had hoped I would become a cancer specialist. While I could not pursue a medical career, I wanted to contribute in my own way. This motivated me to explore how AI and machine learning can help in early cancer detection, making a small difference even outside the medical field.

What it does

AlgoFest_CancerPrediction predicts whether a tumor is Benign or Malignant based on 30 features from a breast cancer dataset. The interactive demo allows:

• Users to input feature values via sliders
  
• Real-time prediction with probabilities
  
• Visual bar charts showing confidence levels for both classes
  
• Colorful, user-friendly interface for easy understanding

How we built it

• Dataset: Breast cancer dataset with 30 features
  
• Models: Logistic Regression, Random Forest, and Decision Tree
  
• Libraries: scikit-learn, pandas, numpy, matplotlib, ipywidgets, joblib
  
• Interface: Jupyter Notebook with ipywidgets for interactive sliders and buttons
  
• Visualization: HTML styling and bar charts for probabilities

Challenges we ran into

• Ensuring realistic predictions instead of defaulting to one class
  
• Handling 30 features with correct value ranges
  
• Creating a clean, interactive demo within Jupyter Notebook
  
• Managing and saving multiple models for consistent predictions

Accomplishments that we're proud of

• Built a full interactive demo with three machine learning models
  
• Successfully implemented real-time probability visualization
  
• Created a system that transforms a personal motivation into a practical AI solution for early cancer detection
  

What we learned

• How to preprocess and handle real-world medical datasets
  
• Training and evaluating multiple ML models using scikit-learn
  
• Building interactive demos with ipywidgets
  
• Visualizing prediction results effectively with HTML and Matplotlib
  

What's next for AlgoFest_CancerPrediction

• Deploy as a web application for broader accessibility
  
• Add support for batch predictions and CSV input
  
• Incorporate additional ML models to improve prediction accuracy
  
• Collect real patient data (anonymized) to fine-tune and validate models further

Built With

• decision
• github
• ipywidgets
• joblib
• joblib-**platform:**-jupyter-notebook-**tools:**-github-(for-version-control-and-project-hosting)-**models-implemented:**-logistic-regression
• jupyternotebook
• logistic-regression
• matplotlib
• numpy
• pandas
• python
• random-forest
• scikit-learn
• seaborn
• tree