assessment

Comparative Analysis of Machine Learning Algorithms for Predicting Heart Failure Outcomes

📌 Overview

This project compares the performance of four machine learning algorithms—Logistic Regression, Support Vector Machine (SVM), K-Nearest Neighbors (KNN), and Gradient Boosting Machine (GBM)—in predicting heart failure outcomes using clinical data.

🧠 Objective

To determine the most effective machine learning model for predicting DEATH_EVENT (mortality) among patients using real-world clinical records.

📂 Project Structure

├── data/ │ └── heart_failure_clinical_records.csv ├── notebooks/ │ └── heart_failure_analysis.ipynb ├── outputs/ │ ├── confusion_matrices/ │ └── classification_reports

📊 Dataset

• Source: UCI Machine Learning Repository
• Records: 299 patients
• Features: 12 clinical variables
• Target Variable: DEATH_EVENT (1 = death, 0 = survived)

🧪 Methodology

1. Data Preprocessing

   • Loaded and cleaned dataset
   • Standardized features for KNN and SVM
   • Train/test split (80/20)

2. Models Trained

   • Logistic Regression
   • SVM (Support Vector Machine)
   • K-Nearest Neighbors (KNN)
   • Gradient Boosting Machine (GBM)

3. Evaluation Metrics

   • Accuracy
   • Precision
   • Recall
   • F1-Score
   • Confusion Matrix

4. Visualization

   • Correlation Matrix
   • Classification Reports
   • Confusion Matrices

🚀 Results

Model	Accuracy	F1-Score	Remarks
Gradient Boosting	✅ Highest	High	Best overall performance
SVM	Moderate	Good	Effective with standardized data
KNN	Moderate	Moderate	Handles non-linear patterns well
Logistic Regression	Moderate	Moderate	Baseline interpretable model

💡 Insights

• Gradient Boosting consistently outperformed others.
• SVM and Logistic Regression are simpler and interpretable.
• KNN works well for non-linear patterns but needs tuning.
• Ensemble models like GBM offer high accuracy for clinical predictions.

📎 Resources

• 📁 Dataset: UCI Repository
• 📄 Report: See MachineLearning_assessment.pdf
• 📦 GitHub Repo: https://github.com/Sravani-Neelakantam/assessment

🛠 Requirements

pandas numpy scikit-learn matplotlib seaborn xgboost 👩‍💻 Author

Sravani Neelakantam MSc Data Science, Coventry University 📧 [email protected]