heart-disease-prediction

Predict heart disease risk using machine learning.

Here's a clean project structure for your README.md (formatted in Markdown):

# Project Structure
heart-disease-prediction/  
├── **data/**  
│   ├── **raw/**               # Raw dataset (`project 2.csv`)
│   └── **processed/**         # Cleaned/preprocessed data (e.g., `cleaned_data.csv`)  
├── **notebooks/**             # Jupyter notebooks for analysis  
│   ├── `eda.ipynb`            # Exploratory Data Analysis  
│   └── `model_training.ipynb` # Model experiments and evaluation  
├── **src/**                   # Python scripts  
│   ├── `data_preprocessing.py` # Data cleaning/feature engineering  
│   ├── `train_model.py`        # Model training and tuning  
│   └── `app.py`                # (Optional) CLI/Flask deployment if we have time
├── **models/**                # Trained models (e.g., `random_forest.pkl`)  
├── **results/**               # Visualizations, metrics, and reports  
│   ├── `confusion_matrix.png`  
│   └── `feature_importance.png`  
├── **requirements.txt**      # Python dependencies  
├── **README.md**             # Project overview and instructions  
└── **OTHERS**               # MIT License (or others if we need any)  

Key Directories Explained

• data/raw: Contains the original unprocessed dataset.
• data/processed: Stores cleaned data after preprocessing.
• notebooks: For exploratory analysis and model prototyping.
• src: Reusable Python scripts for data cleaning, modeling, and deployment.
• models: Saves trained models for later use.
• results: Stores visualizations, performance metrics, and reports.

Collaboration Notes

• Use branches prajith-ravisankar and emilio-santamaria and lasombra7
• Merge changes into dev for daily collaboration after we agree on.
• Final stable code goes into main (protected branch).

Phase 1: Project Setup & Planning (March 13)

Goal: Initialize repository, define roles, and finalize requirements.

• Main Todo 1.1: GitHub Setup
  • Sub-todo 1.1.1: Create a GitHub repository
  • Sub-todo 1.1.2: Creat branches:
    • main: Protected branch for final merges.
    • dev: Shared development branch for daily work.
    • prajith-ravisankar
    • emilio-santamaria teammate has to create their branch and confirm.
    • lasombra7 teammate has to accept the invite from Github and create their branch to start contributions
• Main Todo 1.2: Requirements Finalization
  • Sub-todo 1.2.1: Review the PDF requirements and start working on: - Data cleaning steps (missing values, outliers). - Models to compare (e.g., Logistic Regression, Random Forest, XGBoost). - Metrics (accuracy, F1-score, AUC-ROC).
  • Sub-todo 1.2.2: confirm on communication platform (we are using discord)

Phase 2: Data Preparation (March 14–16)

Goal: Clean and preprocess the dataset.

• Main Todo 2.1: Data Cleaning
  • Sub-todo 2.1.1: Load project 2.csv and inspect for:
    • Missing values (e.g., empty Cholesterol or Blood Pressure entries).
    • Duplicate rows.
    • what to do with outliers? not sure…
      • (e.g., Age > 100, Blood Pressure > 200).
    • etc…
  • Sub-todo 2.1.2: Handle missing values:
    • Use KNNImputer or IterativeImputer for advanced imputation
  • Sub-todo 2.1.3: Encode categorical variables:
    • Gender: Male=0, Female=1.
• Main Todo 2.2: Feature Engineering
  • Sub-todo 2.2.1: Split data into features (X) and target (y).
  • Sub-todo 2.2.2: Scale numerical features:
    • Use StandardScaler (Z-score normalization) for algorithms like SVM or Logistic Regression.
  • Sub-todo 2.2.3: Feature selection:
    • Use SelectKBest or RFE (Recursive Feature Elimination) to reduce dimensionality.
  • Sub-todo 2.2.4: Save preprocessed data as cleaned_data.csv.

---

Phase 3: Exploratory Data Analysis (March 16–17)

Goal: Generate insights and visualizations.

• Main Todo 3.1: Univariate Analysis
  • Sub-todo 3.1.1: Plot distributions for:
    • Age, Cholesterol, Blood Pressure (histograms).
    • Heart Disease (pie chart for class balance).
  • Sub-todo 3.1.2: Document observations (e.g., "30% of patients have heart disease").
• Main Todo 3.2: Bivariate/Multivariate Analysis
  • Sub-todo 3.2.1: Correlation heatmap (features vs. Heart Disease).
  • Sub-todo 3.2.2: Boxplots for Cholesterol vs. Heart Disease.
  • Sub-todo 3.2.3: Pairplot for key features (e.g., Age, Blood Pressure).

---

Phase 4: Model Development (March 18–20)

Goal: Train and compare baseline models.

• [x] Main Todo 4.1: Model Training
  • Sub-todo 4.1.1: Train 3 models:
    • Logistic Regression.
    • Random Forest.
    • XGBoost.
  • Sub-todo 4.1.2: Use train_test_split (80-20 split).
• Main Todo 4.2: Baseline Evaluation
  • Sub-todo 4.2.1: Calculate metrics:
    • Accuracy, Precision, Recall, F1-score, AUC-ROC.
  • Sub-todo 4.2.2: Document results in a shared spreadsheet.

---

Phase 5: Model Optimization (March 21–22)

Goal: Improve model performance.

• Main Todo 5.1: Hyperparameter Tuning
  • Sub-todo 5.1.1: Use GridSearchCV or RandomizedSearchCV for:
    • Random Forest (tune n_estimators, max_depth).
    • XGBoost (tune learning_rate, max_depth).
  • Sub-todo 5.1.2: Re-evaluate metrics post-tuning.
• Main Todo 5.2: Feature Importance
  • Sub-todo 5.2.1: Plot feature importance for the best model.
  • Sub-todo 5.2.2: Identify top 5 risk factors (e.g., Cholesterol, Age).

---

Phase 6: Deployment & Documentation (March 23)

Goal: Prepare a simple deployment and final report.

• Main Todo 6.1: Deployment
  • Sub-todo 6.1.1: Create a predict() function for new data.
  • Sub-todo 6.1.2: Build a basic CLI or Flask app for predictions.