The aim/goal of this project is to build a supervised prediction that can be able to classify whether a patient is prone to Heart Disease or not using the variables collected as shown below
age: Age of the patient (in years)sex: Gender of the patient (0 = female, 1 = male).cp: Chest pain type (0 = typical angina, 1 = atypical angina, 2 = non-anginal pain, 3 = asymptomatic).trestbps: Resting blood pressure (in mm Hg).chol: Serum cholesterol level (in mg/dl).fbs: Fasting blood sugar > 120 mg/dl (0 = false, 1 = true).restecg: Resting electrocardiographic results (0 = normal, 1 = having ST-T wave abnormality, 2 = showing probable or definite left ventricular hypertrophy).thalach: Maximum heart rate achieved.exang: Exercise induced angina (0 = no, 1 = yes).oldpeak: ST depression induced by exercise relative to rest.slope: Slope of the peak exercise ST segment (0 = upsloping, 1 = flat, 2 = downsloping).ca: Number of major vessels colored by fluoroscopy (0-3).thal: Thalassemia (0 = normal, 1 = fixed defect, 2 = reversable defect).target: Presence of heart disease (0 = no, 1 = yes).
This data doesn't need a lot of cleaning like dealing with null values and duplicates, but there is need to evaluate the distribution of the dataset. Check the link for the dataset distribution here The biased variables was standardized using z-score normalization.
The model was trained on XGBoost classifier algorithm. K-fold split was used to train the model on different section of the dataset with their respective accuracy level. I used AUC and confusion matrix as the preferred evaluation metrics.