Novel Approach to Evaluate Classification Algorithms and Feature Selection Filter Algorithms Using Medical Data

Fawad Masood; Junaid Masood; Hina Zahir; Kaouthar Driss; Nasir Mehmood; Hassan Farooq

doi:10.47852/bonviewJCCE2202238

Authors

Fawad Masood School of Computing, Edinburgh Napier University, UK and College of Information Engineering, Yangzhou University, China https://orcid.org/0000-0002-5788-4228
Junaid Masood Department of Computer Science, IQRA National University, Pakistan https://orcid.org/0000-0003-3811-4698
Hina Zahir Department of Electrical Engineering, University of Engineering and Technology, Pakistan
Kaouthar Driss National School of Computer Sciences, ENSI Manouba University Campus, Tunisia
Nasir Mehmood Department of Basic and Applied Sciences, Air University, Pakistan
Hassan Farooq Department of Computer Science, Bahria University, Pakistan

DOI:

https://doi.org/10.47852/bonviewJCCE2202238

Keywords:

data mining, classification algorithm, Tanagra, feature selection algorithm, error rate, input parameters, target parameter

Abstract

In today’s world, hepatitis is a widespread problem related to the medical field, which directly affects the lives of mankind. For patient survival, data mining is essential in predicting future trends using various techniques. This paper uses three feature selection filter algorithms (FSFAs): relief filter, step disc filter, and Fisher filter algorithm and 15 classifiers using a free data mining Tanagra software having UCI Machine Learning Repository. This process is done on a medical dataset with 20 attributes and 155 instances. As a result, the error rate is obtained in terms of accuracy, which shows the performance of algorithms regarding patient survival. This work also shows the independent comparison of FSFAs with classification algorithms using continuous values and the FSFA without using classification algorithms. This paper shows that the obtained result of the classification algorithm gives promising results in terms of error rate and accuracy.

Received: 6 April 2022 | Revised: 12 May 2022 | Accepted: 16 May 2022

Conflicts of Interest

The authors declare that they have no conflicts of interest to this work.