Classification Framework Based on C4.5 Algorithm For Medicinal Data

• Published in: International journal of computer science and information security Vol. 13; no. 4; p. 63
• Main Author: Ganesan, Karthik
• Format: Journal Article
• Language: English
• Published: Pittsburgh L J S Publishing 01.04.2015
• Subjects: Accuracy; Algorithms; Back propagation; Cardiovascular disease; Classification; Constants; Data mining; Datasets; Decision making; Decision trees; Discretization; Heart; Heart diseases; Inference; Manuals; Medical diagnosis; Medical research; Neural networks; Patients; Principal components analysis; Cleveland Ohio
• ISSN: 1947-5500

This study proposes a framework with preprocessing techniques namely Missing value replacement, Discretization, Principal Component Analysis (PCA) to extract the key features and then applying c4.5 classifier algorithm to enhance the classification of medicinal data. The input data gets subjected to missing data imputation through any one of the standard methods like mean, mode, constant and manual input. The dataset is then subjected to Discretization to formalize a reasonable set of discrete bins. PCA is then applied on the dataset to identify the principal components of the dataset, which attribute to the mean data inference. C4.5 algorithm has been used to construct a decision tree based on the information gain of the training set. This work used Cleveland heart disease dataset, obtained from UCI machine learning repository. The dataset is composed of details of about 303 patients and helps to predict presence or absence of cardio vascular disorder based on 75 attributes. The proposed framework was applied on this dataset and exhibited an accuracy of about 77.73%.