Open Access

ARTICLE

Heterogeneous Ensemble Feature Selection Model (HEFSM) for Big Data Analytics

M. Priyadharsini^1,*, K. Karuppasamy²

1 Department of Information Technology, Sri Ramakrishna Engineering College, Coimbatore, Tamilnadu, India
2 Department of Computer Science & Engineering, RVS College of Engineering and Technology, Coimbatore, Tamilnadu, India

* Corresponding Author: M. Priyadharsini. Email:

Computer Systems Science and Engineering 2023, 45(2), 2187-2205. https://doi.org/10.32604/csse.2023.031115

Received 11 April 2022; Accepted 13 June 2022; Issue published 03 November 2022

Abstract

Big Data applications face different types of complexities in classifications. Cleaning and purifying data by eliminating irrelevant or redundant data for big data applications becomes a complex operation while attempting to maintain discriminative features in processed data. The existing scheme has many disadvantages including continuity in training, more samples and training time in feature selections and increased classification execution times. Recently ensemble methods have made a mark in classification tasks as combine multiple results into a single representation. When comparing to a single model, this technique offers for improved prediction. Ensemble based feature selections parallel multiple expert’s judgments on a single topic. The major goal of this research is to suggest HEFSM (Heterogeneous Ensemble Feature Selection Model), a hybrid approach that combines multiple algorithms. The major goal of this research is to suggest HEFSM (Heterogeneous Ensemble Feature Selection Model), a hybrid approach that combines multiple algorithms. Further, individual outputs produced by methods producing subsets of features or rankings or voting are also combined in this work. KNN (K-Nearest Neighbor) classifier is used to classify the big dataset obtained from the ensemble learning approach. The results found of the study have been good, proving the proposed model’s efficiency in classifications in terms of the performance metrics like precision, recall, F-measure and accuracy used.

---

Keywords

---