Heart Disease Prediction Dataset Analysis

Overview

This repository contains the analysis and findings based on the Heart Failure Prediction Dataset. The dataset, sourced from Kaggle, includes clinical information on 918 participants. It features 11 attributes and a binary label, HeartDisease, indicating the presence or absence of heart disease.

The project aimed to:

1. Preprocess and clean the dataset for accurate analysis.
2. Perform exploratory data analysis (EDA) to understand the features and distributions.
3. Apply statistical methods, including hypothesis testing.
4. Build and evaluate logistic regression models for heart disease prediction.

Main Results

• Data Cleaning: Missing values in critical features like cholesterol were imputed using KNN imputation, and original values were preserved for analysis.
• Feature Engineering:
  • One-hot encoding and ordinal encoding were applied to categorical variables.
  • Principal Component Analysis (PCA) addressed multicollinearity issues.
• Statistical Analysis:
  • ANOVA showed no significant differences in cholesterol levels across chest pain types.
  • Pairwise proportion tests revealed significant associations between binary features and heart disease.
• Model Development:
  • Logistic regression models were trained on various feature sets.
  • The best-performing model achieved improved metrics after removing influential outliers.

Key features for heart disease prediction:

• Sex, FastingBS, ExerciseAngina, Oldpeak, ST_Slope, ChestPainType, MissingCholesterolNum.

Performance Metrics:

• High accuracy across all models.
• Receiver Operating Characteristic (ROC) curves showed superior threshold performance for the pruned model.

Model Performance Comparison

	Accuracy	Precision	Recall
Entire Features	0.8696	0.9000	0.8839
Pruned Features	0.8696	0.9000	0.8839
Pruned + No Outliers	0.8587	0.8981	0.8661

Precision-Recall Curve

ROC Curve

Binary Feature Statistical Tests

The following table summarizes results from pairwise proportion z-tests. Significant differences in proportions between classes (based on HeartDisease) are reported, including confidence intervals and effect size estimates.

Group	z-stat	p-value	Reject H₀	Estimator	95% CI Lower	95% CI Upper	p̂₀	p̂₁
Sex	9.24	2.47e-20	✅	0.372	0.301	0.443	0.631	0.259
FastingBS	-8.12	4.84e-16	✅	-0.315	-0.381	-0.249	0.479	0.794
ExerciseAngina	-15.00	6.87e-51	✅	-0.502	-0.556	-0.448	0.350	0.852
ChestPainType_ATA	12.16	4.85e-34	✅	0.510	0.449	0.572	0.649	0.139
ChestPainType_NAP	6.52	7.03e-11	✅	0.258	0.183	0.333	0.610	0.351
ChestPainType_TA	1.65	9.83e-02	❌	0.124	-0.023	0.271	0.559	0.435
RestingECG_LVH	-0.34	7.35e-01	❌	-0.014	-0.093	0.066	0.550	0.564
RestingECG_ST	-3.12	1.80e-03	✅	-0.130	-0.208	-0.051	0.528	0.657
MissingCholesterolNum	-9.63	6.14e-22	✅	-0.406	-0.466	-0.346	0.477	0.883

Getting Started

Prerequisites

Before you begin, ensure that you have the following installed on your system:

• Python (version 3.7 or higher is recommended). We suggest using Anaconda for easier environment and package management.
• Basic knowledge of working with Python virtual environments.

Installation

Follow these steps to set up the repository for analysis:

1. Clone the Repository: Open a terminal and run:

   git clone https://github.com/rdgbrian/heart-data-analysis.git
   cd heart-data-analysis

2. Create a Python Environment: If you are using Anaconda, create and activate a virtual environment:

   conda create --name <env_name> python=3.9
   conda activate <env_name>

   Alternatively, using venv:

   python -m venv <env_name>
   source <env_name>/bin/activate   # On MacOS/Linux
   <env_name>\Scripts\activate      # On Windows

3. Install Dependencies: Use the requirements.txt file to install the necessary packages:

   pip install -r requirements.txt

4. Verify the Setup: To ensure all dependencies are installed correctly, you can test the installation by running:

   python -m pip check

Now you’re ready to begin analysis with the repository!

Usage

To explore and clean the dataset, use the Jupyter notebook:

clean_data_exploratory.ipynb

This notebook performs data preprocessing, exploratory data analysis (EDA), and statistical testing.

modeling.py

This script handles model training, evaluation, and saves performance plots such as the ROC and Precision-Recall curves.

Dataset

The dataset is available here.

Features

Feature	Description
Age	Age of the patient [years]
Sex	Gender of the patient [M: Male, F: Female]
ChestPainType	Chest pain type [TA, ATA, NAP, ASY]
RestingBP	Resting blood pressure [mm Hg]
Cholesterol	Serum cholesterol [mg/dl]
FastingBS	Fasting blood sugar [1: > 120 mg/dl, 0: ≤ 120]
RestingECG	Resting ECG results [Normal, ST, LVH]
MaxHR	Maximum heart rate achieved
ExerciseAngina	Exercise-induced angina [Y: Yes, N: No]
Oldpeak	ST depression induced by exercise
ST_Slope	Slope of the peak exercise ST segment [Up, Flat, Down]

Contributions

Contributions to improve the code, analysis, or extend the findings are welcome. Please follow standard practices for submitting pull requests.

Project Structure

heart-data-analysis/
├── .gitignore                    # Specifies intentionally untracked files to ignore
├── clean_data_exploratory.ipynb # Jupyter notebook for initial data exploration
├── figures.zip                  # Compressed archive of all generated figures
├── modeling.py                  # Main script for training and evaluating ML models
├── model_utils.py               # Helper functions for modeling workflow
├── readme.md                    # Project description, usage, and documentation
├── requirements.txt             # Python package dependencies
├── results.txt                  # Output metrics or experiment results
├── STA_Presentation.pptx       # Project summary presentation (PowerPoint)
│
├── data/                        # Source and processed datasets
│   ├── clean_heart.csv
│   ├── clean_heart_pca.csv
│   ├── heart.csv
│   ├── LifetimeDataANOVA.csv
│   └── MultRegData.csv
│
├── figures/                     # Generated plots and visualizations
│   (figures omitted for brevity)
│
└── rcode/                       # R scripts for statistical analysis
    ├── exploratory2.R
    ├── exploritory.R
    └── MultRegAprt2andANOVAPart1.R

---

License

This project is licensed under the MIT License.