Machine Learning Predictive Model — Heart Disease Detection
Machine learning project comparing classification algorithms for heart disease prediction.
Overview
Situation & Task: Cardiovascular disease accounts for 32% of global mortality. Designed and executed an end-to-end predictive modelling study using the 303-patient Cleveland Heart Disease Dataset (UCI) to identify the most effective ML classification algorithm for early cardiac detection.
Action: Performed comprehensive preprocessing (missing values, duplicates, normalisation) using Python, Pandas, and NumPy across 14 clinical features. Conducted EDA with Matplotlib and Seaborn correlation heatmaps; trained and rigorously evaluated 7 algorithms — Logistic Regression, SVM, KNN, Decision Tree, Random Forest, Gradient Boosting, and Naïve Bayes — using precision, recall, F1-score, and confusion matrices.
Result: Achieved 83% predictive accuracy with the Random Forest model, outperforming all 6 alternatives. Delivered a validated, evidence-based prediction framework applicable to clinical decision support systems.