ZimaBlue01/eastern-cape-health-audit

Health data quality audit and risk analysis for under-resourced clinics. 🏥 Eastern Cape Health Audit Data Quality, Risk Classification & Ethical Analytics The Eastern Cape Department of Health initiated a province-wide audit to improve data quality, chronic illness tracking, and triage decision-making in under-resourced clinics. This project analyses a sample of clinic patient records collected from Mthatha, Queenstown, and rural areas around Lusikisiki, transforming messy health data into clean, interpretable, and ethically-aware insights. 🎯 Objectives This audit focused on five core goals: Clean and standardise patient health data Compare a random clinic sample to the full population Identify demographic patterns using frequency analysis Apply a simple, interpretable Naïve Bayes–style classification Produce a cleaned, classified dataset suitable for policy and triage support 📊 Raw Data (Before Cleaning) The original dataset suffered from: Duplicated records Missing values Inconsistent formatting Wide variation in health metrics Raw Dataset Snapshot Why this matters: Decisions made on unclean health data can lead to misclassification, bias, and unsafe triage outcomes. 🧹 Data Cleaning & Normalisation (After) The dataset was cleaned by: Removing duplicate rows Filling missing numeric values using column means Normalising key health indicators (BMI, blood pressure, disease score) Cleaned Dataset Snapshot This step ensures: Fair comparisons between patients Stable inputs for classification Improved reliability for downstream analysis 📈 Frequency Distribution: Age by Sex To understand demographic structure, a frequency distribution of age by sex was created. Insights: Patients span a wide age range, with clusters in middle-to-older age groups Both male and female patients are represented across most age bands Supports targeted planning for age-related chronic conditions 🔬 Feature Scaling & Preparation Health metrics were normalised to ensure no single variable dominated the classification logic. …