Missing Data Imputation using Machine Learning Algorithm for Supervised Learning

Abstract

With a transience rate of over 18 million per year, Heart Disease (HD) has emerged out to be the lethal disease of the world. Data mining-based heart disease diagnosis systems can surely aid cardiac professionals in a timely diagnosis of the patient's condition. In this proposed work, a Python-based data mining system capable of diagnosing the HD using a Decision Tree has been developed. In the methodology, the UCI data repository was taken into consideration with 14 Attributes. In the dataset, there are few missing values (yet found to be hyperparameter), and pre-processing with such missing values is a common yet challenging problem. A mere substitution will give biased results from the data to be observed for HD diagnosis and will certainly affect the value of the learning process in Machine Learning. Therefore, in the proposed work, a missing value imputation is done, which gave better accuracy, and it is trustable.