South African heart disease analysis with SAS
# ❤️ Coronary Heart Disease Risk Factor Analysis (SAS)
A SAS-based statistical analysis identifying the risk factors most
strongly associated with Coronary Heart Disease (CHD), using the South
African Heart Disease dataset — covering variable identification,
descriptive statistics, outlier treatment, hypothesis testing, and a
logistic regression model with strong predictive performance (AUC = 0.8327).
**SAS coding & analysis:** Md Tariqul Islam (Tariq)
**Presentation & summary:** Haiyan Zhu
---
## 📋 Table of Contents
- Overview
- Dataset
- Conceptual Framework
- Methodology
- Key Findings
- Recommendations
- Repository Structure
- Reproducing the Analysis
- Tech Stack
---
## Overview
Coronary Heart Disease (CHD) is one of the leading causes of death among
adults worldwide. This project analyzes a retrospective dataset of males
from three heart-disease high-risk communities in the Western Cape, South
Africa, to answer two business questions: **what factors increase CHD
risk, and how can the disease be avoided?**
The analysis proceeds from raw variable identification through descriptive
statistics, outlier treatment, hypothesis testing, and finally a logistic
regression model that fits the data well (AUC of ROC = 0.8327), concluding
with concrete prevention recommendations.
📄 Full write-up of the business problem, framework, and findings:
`docs/project_brief.md`
## Dataset
**South African Heart Disease Dataset** (OpenML #1498)
Originally published in: Rousseauw J, et al. *"Coronary risk factor
screening in three rural communities."* South African Medical Journal, 64
(1983), 430–436.
- **462 observations**, 10 columns (2 categorical, 8 numeric variables)
- **Target variable:** `Class` — CHD response (Yes/No)
- **Predictors:** SBP, tobacco use, LDL cholesterol, adiposity, family
history, type-A behavior, obesity, alcohol consumption, age at onset
## Conceptual Framework
```
Y (Class: CHD Yes/No) = f(Xs)
```
| Predictor | Description |
|---|---|
| V1 (SB …