Simplified Novel Approach for Accurate Employee Churn Categorization using MCDM, De-Pareto Principle Approach, and Machine Learning

Aug 5, 2026

Upload by Eliana

Employee churn, reflecting employee turnover within organizations, causes negative effects on productivity, viability, competitiveness, and overall benefits. Manual identification of churn and related employee categorization is difficult, so machine learning methods are commonly used. This work proposes a novel multi-criterion decision-making (MCDM) approach with the DE-Pareto principle, forming the SNEC scheme. An AHPTOPSIS DE-Pareto model performs a two-stage process using AHP for factor weights and TOPSIS for categorization significance, then applies a 20-30-50 DE-Pareto rule to form three groups: enthusiastic, behavioral, and distressed. A Random Forest baseline predicts class-wise churn and is evaluated against other ML methods, achieving similar or slightly better accuracy and MCC with lower time complexity than the ECPR CATBOOST scheme.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages19

Reading Time48 min

AccessFree to download

Introduction

Employee churn problem and impact

Factors influencing employee churn

Related work and gaps

Proposed SNEC framework

Two-stage MCDM with AHP and TOPSIS

DE-Pareto 20-30-50 rule and three employee groups

Experimental setup and evaluation

Baseline Random Forest and dataset

Comparison with other ML methods

Time complexity and performance metrics

| 2024, 21(2 Special Issue): 0706-0724
[https://doi.org/10.21123/bsj.2024.9788](https://doi.org/10.21123/bsj.2024.9788)
[P-ISSN: 2078-8665-E-ISSN: 2411-7986](P-ISSN: 2078-8665-E-ISSN: 2411-7986)

Baghdad Science Journal | | --- | | Simplified Novel Approach for Accurate Employee Churn Categorization using MCDM, De-Pareto Principle Approach, and Machine Learning
Faisal BinAlAbid1, Aryati Binti Bakri2, Md. Golam Rabiul Alam3, Jia Uddin*4
, Shefayatuj Johara Chowdhury5
1Faculty of Computing, Universiti Teknologi Malaysia, Johor Bahru, Malaysia.
2Faculty of Computing, Informatics Engineering, Universiti Teknologi Malaysia, Johor Bahru, Malaysia.
3Department of Computer Science and Engineering, Brac University, Dhaka, Bangladesh.
4Department of AI and Big Data, Endicott College, Woosong University, Daejeon, South Korea.
5Department of Computer Science and Engineering, International Islamic University Chittagong, Chittagong, Bangladesh.
*Corresponding Author.
PARS2023: Postgraduate Annual Research Seminars 2023.
Received 30/09/2023, Revised 10/02/2024, Accepted 12/02/2024, Published 25/02/2024
© 2022 The Author(s) . Published by College of Science for Women, University of Baghdad.
This is an Open Access article distributed under the terms of the Creative Commons Attribution 4.0 International License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. | Abstract Churning of employees from organizations is a serious problem. Turnover or churn of employees within an organization needs to be solved since it has negative impact on the organization. Manual detection of employee churn is quite difficult, so machine learning (ML) algorithms have been frequently used for employee churn detection as well as employee categorization according to turnover. Using Machine learning, only one study looks into the categorization of employees up to date. A novel multi-criterion decision-making approach (MCDM) coupled with DE-PARETO principle has been proposed to categorize employees. This is referred to as SNEC scheme. An AHPTOPSIS DE-PARETO PRINCIPLE model (AHPTOPDE) has been designed that uses 2-stage MCDM scheme for categorizing employees. In 1st stage, analytic hierarchy process (AHP) has been utilized for assigning relative weights for employee accomplishment factors. In second stage, TOPSIS has been used for expressing significance of employees for performing employee categorization. A simple 20-30-50 rule in DE PARETO principle has been applied to categorize employees into three major groups namely enthusiastic, behavioral and distressed employees. Random forest algorithm is then applied as baseline algorithm to the proposed employee churn framework to predict class-wise employee churn which is tested on standard dataset of the (HRIS), the obtained results are evaluated with other ML methods. The Random Forest ML algorithm in SNEC scheme has similar or slightly better overall accuracy and MCC with significant less time complexity compared with that of ECPR scheme using CATBOOST algorithm. Keywords: AHP-TOPSIS, DE-PARETO principle, Employee churn, MCDM, Random Forest algorithm. Introduction What is employee churn? What happens within an organization when employees churn? The departure of individuals and subsequently intellectual capital from an organization is known as employee churn 2024, 21(2 Special Issue): 0706-0724 [https://doi.org/10.21123/bsj.2024.9788](https://doi.org/10.21123/bsj.2024.9788) [P-ISSN: 2078-8665-E-ISSN: 2411-7986](P-ISSN: 2078-8665-E-ISSN: 2411-7986) Baghdad Science Journal or turnover 1. Employee churn can have an adverse influence on the productivity, viability, competitiveness, and benefit within an organization2. In this context, preventing employee churn within an organization is essential for its existence. This problem of employee turnover has become a persistent phenomenon f

FAQs

What approach does the paper use to categorize employee churn?

It proposes the SNEC scheme, a multi-criterion decision-making approach coupled with the DE-Pareto principle to categorize employees into enthusiastic, behavioral, and distressed groups.

How are weights and significance determined in the proposed model?

The first stage uses AHP to assign relative weights for employee accomplishment factors, and the second stage uses TOPSIS to express the significance of employees for categorization.

How is the churn prediction performance evaluated?

Random Forest is applied as a baseline to predict class-wise employee churn, and results are evaluated against other machine learning methods using metrics such as overall accuracy and MCC while also comparing time complexity.

Aug 5, 2026

Upload by Eliana