Heart Disease Prediction With SVM

Heart Disease Data

• Data from three hospital systems: Cleveland, Hungary, Switzerland
• This database contains 76 attributes, but all published experiments refer to a subset of 14 of them
• The goal field refers to the presence of heart disease in the patient, an integer valued from 0 (no presence) to 4
• Names and social security numbers of patients were removed from the database, replaced with dummy values

Heart Disease Dataset

• 13 Independent Variables
• Prediction Output (0-4)

Implementation

• Missing value handled using SimpleImputer
• Splited test data and train data
• StandardScaleris used to fix class imbalance
• Support Vector Classification(SVC) is used and kernel used is rbf as others(linear, poly, sigmoid) give less accuracy