Eksperimen SML Muhammad Rizki

Template Eksperimen Machine Learning untuk Klasifikasi Penyakit Jantung menggunakan Heart Failure Prediction Dataset dari Kaggle.

📊 Dataset Information

• Dataset: Heart Failure Prediction Dataset
• Source: Kaggle - Heart Failure Prediction
• Size: 918 samples, 12 features
• Task: Binary Classification (Heart Disease: 0/1)
• Features: Age, Sex, ChestPainType, RestingBP, Cholesterol, FastingBS, RestingECG, MaxHR, ExerciseAngina, Oldpeak, ST_Slope, HeartDisease

🏗️ Repository Structure

Eksperimen_SML_[Nama-siswa]/
├── .github/
│   └── workflows/
│       └── preprocessing-workflow.yml    # GitHub Actions workflow
├── heart-failure-prediction/                           # Raw dataset directory
│   └── heart.csv
├── preprocessing/
│   ├── Eksperimen_[Nama-siswa].ipynb   # Experimentation notebook
│   ├── automate_[Nama-siswa].py        # Automated preprocessing script
│   └── heart_preprocessing/            # Processed dataset directory
│       ├── X_train.csv
│       ├── X_test.csv
│       ├── y_train.csv
│       ├── y_test.csv
│       └── heart_processed.csv
├── requirements.txt                     # Python dependencies
├── README.md                           # This file
└── preprocessing_report.md             # Automated preprocessing report

🎯 Kriteria Submission

✅ Basic (2 pts)

• Melakukan tahapan experimentation secara manual
• Data loading pada notebook
• EDA (Exploratory Data Analysis) pada notebook
• Preprocessing pada notebook

✅ Skilled (3 pts)

• Semua tahap Basic terpenuhi
• File automate_[Nama-siswa].py dengan fungsi preprocessing otomatis
• Konversi dari proses eksperimen dengan struktur berbeda
• Mengembalikan data siap latih

✅ Advance (4 pts)

• Semua tahap Skilled terpenuhi
• GitHub Actions workflow untuk preprocessing otomatis
• Repository dengan struktur folder sesuai kriteria
• Actions mengembalikan dataset terproses

🚀 Quick Start

1. Setup Environment

# Clone repository
git clone https://github.com/username/Eksperimen_SML_[Nama-siswa].git
cd Eksperimen_SML_[Nama-siswa]

# Install dependencies
pip install -r requirements.txt

2. Manual Experimentation

Buka dan jalankan notebook preprocessing/Eksperimen_[Nama-siswa].ipynb di Jupyter atau Google Colab:

jupyter notebook preprocessing/Eksperimen_[Nama-siswa].ipynb

3. Automated Preprocessing

from preprocessing.automate_[Nama-siswa] import preprocess_heart_disease_data

# Run automated preprocessing
X_train, X_test, y_train, y_test, preprocessor = preprocess_heart_disease_data('heart.csv')

# Use preprocessed data for model training
print(f"Training data shape: {X_train.shape}")
print(f"Testing data shape: {X_test.shape}")

4. GitHub Actions (Automatic)

Workflow akan otomatis berjalan ketika:

• Push ke branch main atau master
• Ada perubahan pada folder preprocessing/ atau file automate_*.py
• Manual trigger melalui GitHub Actions tab
• dataset hasil dapat diakses pada artifact workflow

📝 Preprocessing Steps

1. Data Loading

• Load dataset dari CSV file
• Validasi struktur dan tipe data

2. Missing Value Handling

• Identifikasi missing values
• Handle nilai 0 yang tidak valid pada Cholesterol dan RestingBP
• Imputasi dengan median untuk nilai numerik

3. Duplicate Removal

• Deteksi dan hapus data duplikat
• Preserve data integrity

4. Categorical Encoding

• One-Hot Encoding: ChestPainType, RestingECG, ST_Slope
• Label Encoding: Sex, ExerciseAngina (binary variables)

5. Feature Scaling

• StandardScaler untuk fitur numerik: Age, RestingBP, Cholesterol, MaxHR, Oldpeak
• Preserve distribusi data

6. Train-Test Split

• Ratio: 80% training, 20% testing
• Stratified split untuk menjaga distribusi target
• Random state: 42 (reproducible)

📊 EDA Highlights

Dataset Overview

• Total Samples: 918
• Features: 11 input features + 1 target
• Target Distribution: ~55% Heart Disease, ~45% No Heart Disease
• Missing Values: None (explicit)
• Data Quality Issues: Zero values in Cholesterol (172 samples)

Key Insights

1. Age Distribution: Most patients between 50-60 years
2. Gender: More male patients (79%) than female (21%)
3. Chest Pain: ASY (Asymptomatic) most common type
4. Heart Rate: Lower max HR associated with heart disease
5. Exercise Angina: Strong predictor of heart disease

🤖 GitHub Actions Workflow

Triggers

• Push to main/master branch
• Changes in preprocessing files
• Manual workflow dispatch

Steps

1. Environment Setup: Python 3.9, dependencies installation
2. Data Download: Kaggle API integration (optional)
3. Preprocessing: Run automated pipeline
4. Validation: Quality checks on processed data
5. Artifacts: Upload processed files
6. Reporting: Generate preprocessing report

📋 Usage Examples

Using Preprocessor Class

from preprocessing.automate_[Nama-siswa] import HeartDiseasePreprocessor

# Initialize preprocessor
preprocessor = HeartDiseasePreprocessor()

# Load and preprocess data
df = preprocessor.load_data('heart.csv')
X_train, X_test, y_train, y_test = preprocessor.fit_transform(df)

# Transform new data
new_data_processed = preprocessor.transform(new_data)

# Save processed data
preprocessor.save_processed_data(X_train, X_test, y_train, y_test, './output/')

Direct Function Call

from preprocessing.automate_[Nama-siswa] import preprocess_heart_disease_data

# One-line preprocessing
result = preprocess_heart_disease_data('heart.csv', './output/')
X_train, X_test, y_train, y_test, preprocessor = result

🔧 Customization

Modifying Preprocessing Steps

Edit automate_[Nama-siswa].py to customize:

• Feature selection
• Encoding strategies
• Scaling methods
• Train-test split ratio

Adding New Features

1. Update HeartDiseasePreprocessor class
2. Modify column definitions
3. Test with sample data
4. Update documentation

📊 Model Training Example

from sklearn.ensemble import RandomForestClassifier
from sklearn.metrics import classification_report

# Load preprocessed data
X_train, X_test, y_train, y_test, _ = preprocess_heart_disease_data('heart.csv')

# Train model
model = RandomForestClassifier(n_estimators=100, random_state=42)
model.fit(X_train, y_train)

# Evaluate
y_pred = model.predict(X_test)
print(classification_report(y_test, y_pred))

🐛 Troubleshooting

Common Issues

1. Kaggle API Error

   # Check API credentials
   cat ~/.kaggle/kaggle.json
   # Re-download credentials from Kaggle

2. Missing Dependencies

   pip install -r requirements.txt

3. File Path Issues

   # Use absolute paths
   import os
   file_path = os.path.abspath('heart.csv')

GitHub Actions Issues

1. Workflow Not Triggering

   • Check file paths in trigger conditions
   • Verify branch names (main vs master)

2. Kaggle Download Fails

   • Add KAGGLE_USERNAME and KAGGLE_KEY secrets
   • Check dataset URL

📚 References

• Heart Failure Prediction Dataset
• Scikit-learn Documentation
• GitHub Actions Documentation
• Template MSML Guidelines

👥 Contributing

1. Fork the repository
2. Create feature branch (git checkout -b feature/improvement)
3. Commit changes (git commit -am 'Add improvement')
4. Push to branch (git push origin feature/improvement)
5. Create Pull Request

---

⭐ Star this repository if it helped you!