Title: An assessment of classification with hybrid methodology for neural network classifier against different classifier

Authors: Aakanksha Jain; Abhishek Kumar; Jyotir Moy Chatterjee; Pramod Singh Rathore

Addresses: Shri Ratan lal Kanwarlal Patni Girls College Kishangarh, Ajmer, Rajasthan-305801, India ' Aryabhatta College of Engineering and Research Center, Ajmer, Rajsthan-302028, India ' Lord Buddha Education Foundation (APUTI), Kathmandu 44600, Nepal ' Aryabhatta College of Engineering and Research Center, Ajmer, Rajsthan-302028, India

Abstract: This research is an assessment of classification with neural network classifier (NNC) against various classifiers centred on working effectiveness of various classifiers. We have compared resultant factors of NBC with other two algorithms, namely ripple down rule learner (RIDOR) and simple cart in order to get comparatively efficient and accurate results. NBC performs well on categorical as well as on numerical data. Along these lines, we have proposed a model of a hybrid technology of analysing accuracy proportion of NNC with NBC, rule-based classifiers and tree-based classifiers on diagnosis of heart disease dataset. Algorithms which we have used here are NBC, ripple down rule learner (RIDOR) and simple cart. This work considered substantial dataset and distinctive methodology for the connected classifier work NBC calculation, is taken as base methodology and every methodology like RIDOR and simple cart are utilized for examination, so as to anticipate heart disease status of patients.

Keywords: naive Bayes classifier; NBC; trees classifier; simple cart; rule classification; RIDOR; neural network classifier; NNC; hybrid methodology; classifier; data mining; heart disease.

DOI: 10.1504/IJCI.2020.111659

International Journal of Collaborative Intelligence, 2020 Vol.2 No.2, pp.83 - 93

Received: 19 Dec 2018
Accepted: 22 Feb 2019
Published online: 08 Dec 2020 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article