Fuzzy Based Model for Mitigating Employee Attrition

Nida Hasib

, Syed Wajahat Abbas Rizvi

and Vinodani Katiyar

Amity University Uttar Pradesh, India

DSMNR University, Lucknow, India

Keywords: Model for Mitigating Employee Attrition, MMEA, Risk Code, Risk Metrics, Fuzzy Inference System, FIS.

Abstract: Employee attrition is a major concern for IT firms in today's corporate environment. Aside from the loss of

human resources, employee turnover also diminishes the organization's ability to use the expertise and

revenue-generating potential of those individuals. This study proposes a fuzzy logic-based phase-wise Model

for Mitigating Employee Attrition (MMEA) that evaluates employee attrition at each stage of the software

development life cycle using the most pertinent risk measures. The research has made use of the fuzzy

inference process power in creating a model based on the anticipated and reduced staff attrition. Using data

from sixteen actual software projects, the suggested model's predictive accuracy is confirmed. The MMEA

model developed as per the guidelines of the proposed framework may help software professionals to take

appropriate corrective measures to predict and reduce employee attrition during software development life

cycle for efficient and accurate software development process in IT sector. By giving management of the

organization the ability to proactively address attrition-related issues and make long-term strategic decisions

that benefit the company, the model effectively maximizes staff retention, according to the research. Our

results produced proof that the alternate strategy was valid. As a result, managers and companies may find a

more practical tool in the used method for evaluating employee decline.

1 INTRODUCTION

Employee attrition is a significant concern for

organizations, leading to substantial losses in IT

industry. Each organization’s context is unique, so

tailored strategies are essential. Employee attrition in

the IT industry can be influenced by several factors,

including absenteeism, performance, and

engagement. By combining predictive models, data

insights, and employee development, companies can

effectively reduce attrition rates in the software

development industry (Hasib et al. 2023).

There are various number of techniques through

which employee attrition can be mitigated in software

development industry- Data Analytics and Insights,

Upskilling and Empowering Managers, Predictive

Models Using Machine Learning, Fuzzy logic. Fuzzy

is the term used to describe things that are ambiguous

or unclear. Fuzzy is the term used to describe the

things that we commonly encounter in the real world

that are ambiguous or confusing. Fuzzy logic

https://orcid.org/0000-0001-8178-422X

https://orcid.org/0009-0006-8064-9388

provides tremendously helpful thinking flexibility

since we often encounter circumstances in the real

world when we are unable to determine whether a

condition is true or untrue.

Figure 1: Fuzzy Logic Architecture.

Fuzzy logic is created using fuzzy rules, which are if-

then statements that depict the relationship between

input and output variables in a fuzzy way as rule base.

A fuzzy logic system produces a fuzzy set, which is a

collection of membership degrees for each possible

Hasib, N., Rizvi, S. W. A. and Katiyar, V.

Fuzzy Based Model for Mitigating Employee Attrition.

DOI: 10.5220/0013460500003956

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 7th International Conference on Finance, Economics, Management and IT Business (FEMIB 2025), pages 105-117

ISBN: 978-989-758-748-1; ISSN: 2184-5891

105

output value

(Yadav and Yadav, 2015), (Nikmanesh,

2023).

Figure 1 depicts a fuzzy logic architecture that

handles fuzzification and defuzzification.

Figure 2 depicts a comprehensive fuzzy logic

system for reliability modelling. Fuzzy logic systems

consist of four primary parts: fuzzy rule base, fuzzy

inference process, fuzzy membership function

(input), and defuzzification (output). The process of

converting a clear value into a fuzzy value is called

fuzzification. The input and output variables are

fuzzified using linguistic variables such as low (L),

medium (M), and high (H) based on the available data

and related uncertainty. The fuzzy rule base is the

fundamental building block of all fuzzy systems.

Figure 2: Overview of Fuzzy Logic System for mitigating

employee attrition.

The fuzzy rule foundation is made up of historical

data, human knowledge expertise, and failure

analysis. These rules are implemented in an

acceptable and effective manner using the other

fuzzy system components.

To sum up, fuzzy logic is a mathematical

framework that captures ambiguity and uncertainty in

decision-making; it has many uses and permits partial

truths. There is an intermediate value in fuzzy logic,

nevertheless, that is both partially true and partially

false. Thus, utilizing the risk measures that impact

employee attrition throughout the SDLC phases, a

fuzzy logic based phase-wise employee attrition

recognition and mitigation model is presented in this

article (Yadav and Yadav, 2015).

The rest of the paper is organized as follows: In

section 2, related work is discussed. In section 3, the

proposed framework is presented. Section 4 describe

implementation of phases of model. Section 5

describe empirical validation of sixteen case studies

and predicted result of MMEA, Section 6 and 7

predictive accuracies of MMEA and quantitative

comparison with other models. Conclusion and future

extensions are presented in section 8.

2 FRAMEWORK FOR

MITIGATING EMPLOYEE

ATTRITION

In continuation with the highlighted need and

significance as discussed in previous section, the

researcher has already proposed a structured

framework for Mitigating Employee Attrition (Figure

2.) based on biological immune system based

artificial immune system conceptual theory as a

solution for the identified inadequacies present in

earlier employee attrition evaluation studies (Hasib et

al., 2023) (Hasib et al., 2024).

The framework described a comprehensive

employee attrition quantification process through its

eight phases (Conceptualization, Initialization and

Recognition, Correlation and Association,

Development and Quantification, Analysis and

Finalization) as depicted in Figure 3. It has been

designed in such a way that both industry personnel

and researchers will find it simple to execute. The

framework focuses on all phases of the software

development life cycle. The researcher thoroughly

defined all of the framework's phases, as well as its

key attributes, which support its claim to be a better

employee attrition framework. (Chauhan and Patel,

2013) (Hasib et al., 2024)

Figure 3: Framework for Mitigating Employee Attrition.

3 FRAMEWORK

IMPLEMENTATION

In the proposed mode (MMEA), employee attrition

indicator during all the phases of SDLC using eleven

risk metrics for recognition and mitigation of

FEMIB 2025 - 7th International Conference on Finance, Economics, Management and IT Business

106

employee attrition of IT sectors. The model was

developed using a fuzzy inference technique, and risk

metrics are evaluated in language terms.

3.1 Implementing Conceptualization

Phase

In terms of the framework, this phase serves as the

foundation for the subsequent phases. This is the

initial phase in developing a comprehensive solution

to a problem. The image illustrates two subtasks: The

importance of applying the AIS idea and

implementing a risk mitigation framework during

software development are discussed in (Hasib et al.,

2023) (Hasib et al., 2024). The first two sections of this

paper and past research work covered all three

conceptual subtasks.

3.2 Implementing Initialization and

Recognition Phase

Certain risk issues pose a hazard to every stage of the

SDLC, from the project's first examination to its final

release. The risk factors that are relevant to every stage

of the SDLC are continually changing requirements,

time contention, project funding loss, team attrition,

data loss, miscommunication. One the key factor that

effect software development process most according

to literature review is employee attrition. Employee

attrition effects and disturb continuous processing of

Software development life cycle phases and impact

IT sectors/industries in its cost, efficiency and

productivity.

There are number of causes of employee attrition

in IT industries during software development process.

A number of employee attrition reduction

frameworks uses risk metrics has been proposed in

last two decades. The accuracy of predictions may

rise with the selection of risk measures. But the most

important factors in lowering employee attrition have

to be taken into account. As a result, the researcher

obtained a number of risk measures from various

available sources through a literature study. As per

the comprehensive literature review performed by

researcher there are number of variables exist in

literature that effects employee attrition in IT sectors

(Gupta and Bhatia, 2023), (Rusi and Viollet, 2023).

Researcher has taken top most recommended risk

metrics that effects most out of number of factors

which are reasons of employee attrition in IT industry

(Table 1) (Hasib et al., 2025). Those are considered

as risk metrics for recognition and mitigation of

employee attrition through quantification analysis of

our model (MMEA) (Figure 4). The objective of the

initialization and recognition phase is to initialize and

recognize the effectiveness of factors that are related

directly or indirectly to the employee attrition during

software development process.

Figure 4: Model for Mitigating Employee Attrition.

Table 1: Risk Code and Metrics.

S.NO.

Risk

Code

Risk Metrics

1 DE Decreased Engagement

2 IA Increased Absenteeism

3 DIP Decline in Performance

4EAIEm

ee Attrition Indicator

5 OP-OC

Organization

Culture

(Openness-Organizational

Culture)

6 TR-CG

Career Growth opportunities

(Training-Career Growth

opportunities)

APP-

CRR

Appraisal

(Appraisal-Compensation,

reward, recognition)

FLEX-

WLB

Work Life Balance

(Flexibility-Work Life

Balance)

9 JS-SAT Job Satisfaction

EMP-

SAT

Employee Satisfaction

EMP-

ATT

Employee Attrition

3.3 Implementing Correlation and

Association Phase

In this step of the framework the researcher has

shortlisted eleven metrics out of others form the

literature review of different organization dealing with

evaluation of employee attrition for betterment of

Fuzzy Based Model for Mitigating Employee Attrition

107

software development process in IT industries (Hasib et

al., 2025). Out of these some refer to risk recognition

phase and other refers to risk mitigation (Kermani,

2021). All risk metrics are assigned with linguistic

values after expert renewal. After that correlation and

association process conducted through risk matrix

between key risk metrics (employee attrition indicator

and employee attrition, job satisfaction, employee

satisfaction) with other risk metrics on the basis of if-

then analysis. (Figure 5).

Figure 5: Risk metric analysis in FIS1, FIS2, FIS3, FIS4

using risk matrix.

After rationalizing association and correlation with

respect to employee attrition of recognized risk

metrics in above section, finally is to freeze metric set

which can be mutated according to strategies of

organization environmental condition during

software development process and employee attrition

mitigation accuracy will be maintained.

3.4 Implementing Development and

Quantification Phase

Since the actual development of the MMEA occurs

during this phase, it is the most important one in the

framework. The model is implemented using the

Mandani type-1 fuzzy logic toolbox in MATLAB

R2024a. The model's fundamental steps include

choosing risk metrics (input/output variables),

creating a fuzzy profile of these variables, creating a

fuzzy rule base, and utilizing a fuzzy inference system

(FIS) to recognize and mitigate employee attrition

throughout the software development process at all

stages.

The forms of membership functions can be

polygonal, trapezoidal, triangular, and more. Triangle

membership functions are taken into consideration in

this study for the creation of fuzzy profiles for a

variety of identified input/output variables (Table 2).

Due to its simplicity and ease of comprehension,

triangular membership functions (TMFs) are

frequently employed for the computation and

interpretation of employee attrition statistics.

Table 2: Risk metrics range of membership function.

NO.

Risk

Metrics

Input/ Output

metrics

MF range

(0-1)

1 DE Less, Somewhat,

[-.5 0 .5],

[0 .5 1], [.5

1 1.5]

2 IA Low, Moderate,

Substantial

[-.5 0 .5],

[0 .5 1], [.5

1 1.5]

3 DIP Low, Moderate,

Substantial

[-.5 0 .5],

[0 .5 1], [.5

1 1.5]

4 EAI Low, Medium,

High

[-.5 0 .5],

[0 .5 1],

[.5 1 1.5]

5 OP-OC Poor, Average,

Good

[-.5 0 .5],

[0 .5 1],

[.5 1 1.5]

6 TR-CG Limited, Average,

Ample

[-.5 0 .5],

[0 .5 1],

[.5 1 1.5]

7 APP-CRR Low, Medium,

High

[-.5 0 .5],

[0 .5 1], [.5

1 1.5]

8 FLEX-WLB Poor, Average,

Good

[-.5 0 .5],

[0 .5 1], [.5

1 1.5]

9 JS-SAT Dissatisfied,

slightly satisfied,

neutral,

somewhat

satisfied, satisfied

[-.25 0 .25],

[0 .25 .5],

[.25 .5 .75],

[.5 .75 1],

[.75 1 1.25]

10 EMP-SAT Dissatisfied,

slightly satisfied,

neutral,

somewhat

satisfied, satisfied

[-.25 0 .25],

[0 .25 .5],

[.25 .5 .75],

[.5 .75 1],

[.75 1 1.25]

11 EMP-ATT Very Low, Low,

Medium, High,

Very High

[-.25 0 .25],

[0 .25 .5],

[.25 .5 .75],

[.5 .75 1],

[.75 1 1.25]

Low Moderate Substantial Low Moderate Substantial Low Moderate Substantia

Less Low Medium High Less Low Medium High Low Low Medium High

Some

Low Medium High

Somewhat

Low Medium High

Moderate

Low Medium High

More Medium High High More Medium High High Substantia Medium High High

EAI (Employee Attrition Indicator) EAI (Employee Attrition Indicator) EAI (Employee Attrition Indicator)

Poor Average Good Limited Average Ample Limited Average Ample

Low

Somewhat

satisfied

Somewhat

satisfied

Satisfied

Low

Somewhat

satisfied

Somewh

ti fi d

Satisfied

Poor

Dissatisfie

Slightly

Satisfied

Somewha

t satisfied

Med ium

Slightly

Satisfied

Nuetral Somewha

t satisfied

Med ium

Slightly

Satisfied

Nuetral Somewha

t satisfied

Average

Slightly

Satisfied

Nuetral Somewha

t satisfied

High

Dissatisfied Slightly

Satisfied

Slightly

Satisfied

High

Dissatisfie

Slightly

Satisfied

Slightly

Satisfied

Good

Somewhat

satisfied

Somewh

satisfied

Satisfied

JS-SAT (Job Satisfaction) JS-SAT (Job Satisfaction) JS-SAT (Job Satisfaction)

Poor Average Good Low Medium High Poor Average Good

Low

Dissatisfied Slightly

satisfied

Somewha

t satisfied

Low Somewhat

satisfied

Somewh

satisfied

Satisfied Low Somewhat

satisfied

Somewh

satisfied

Satisfied

Med ium

Slightly

satisfied

Nuetral Somewha

t satisfied

Medium Slightly

satisfied

Nuetral Somewha

t satisfied

Medium Slightly

satisfied

Nuetral Somewha

t satisfied

High

Somewhat

satisfied

Somewhat

satisfied

Satisfied High Dissatisfie

Slightly

satisfied

Slightly

satisfied

High Dissatisfie

Slightly

satisfied

Slightly

satisfied

EMP-SAT (Em ployee Satisfaction) EMP-SAT (Em ployee Satisfaction) EMP-SAT (Employee Satisfaction)

Dissatisfied Slightly

satisfied

Nuetral Somewhat

satisfied

Satisfied

Dissatisfied

VH H M L VL

Slightly

satisfied

HHMLVL

Neutral

MMMLVL

Somewhat

satisfied

LLLVLVL

Satisfied

VL VL VL VL VL

EMP-ATT ( Employee Attrition )

DIP

EAI

TR-CG

EAI

TR-CG

OP-O

FLEX-W

APP-CRR

OP-OC

EAI

APP- CRR

FLEX-WLB

EMP-SAT

JS-SAT

FEMIB 2025 - 7th International Conference on Finance, Economics, Management and IT Business

108

FIS 1

Figure 6: Fuzzy Inference System (FIS 1) plot.

Figure 7: 27 Rules of FIS1.

FIS 2

Figure 8: Fuzzy Inference System (FIS 2) plot.

Figure 9: 27 Rules of FIS2.

FIS 3

Figure 10: Fuzzy Inference System (FIS 3) plot.

Figure 11: 27 Rules of FIS3.

FIS 4

Figure 12: Fuzzy Inference System (FIS 4) plot.

Figure 13: 25 Rules of FIS4.

Fuzzy Based Model for Mitigating Employee Attrition

109

From the above correlation and association phase, it

has been visualized that range of membership

function are created between 0-1. As shown in above

(Table 2) first eight risk metrics have three MFs

ranges and last three risk metrics form the list have

five MFs ranges. On the basis of the previous phase,

we come to know how much fuzzy rules are to be

prepared, 106 fuzzy rules are prepared for dealing

with recognition and mitigation of employee attrition

using different risk metrics. With reference to these

fuzzy rules, fuzzy inferences are reflected with

various significant values from organizations work

modules document. Employee attrition evaluation

performed through these rules and their inferences to

recognize level of attrition and try to reduce attrition

percentage through manipulating various risk metrics

in proposed model based on organization

environmental condition. The explanatory process of

proposed model is shown above in this section as 4

Fuzzy Inference System (FIS1, FIS2, FIS3, FIS4)

which consist of Fuzzy Inference System(FIS) plot;

property editor of all FIS consist of -implication

method (min), aggregation method (max),

defuzzification method (centroid); membership

function plot for every metrics showing degree of

membership; rule editor showing all possible rules

created in every fuzzy inference system; rule

inference system. In proposed model FIS 1 consist of

27 rules, FIS 2 consist of 27 rules, FIS 3 consist of 27

rules, FIS 4 consist of 25 rules. In all total 106 rules

to solve employee attrition mitigation problem in IT

sectors. (Figure 6 to Figure 13) shows all fuzzy

profiles of FIS 1, FIS2, FIS3, FIS4 including its fuzzy

inference system plot, fuzzy profiles with

membership ranges, and fuzzy rules (Ahmed et al.,

2013).

3.5 Implementing Analysis and

Finalization Phase

Although the developed Model for Mitigating

Employee Attrition has been theoretically and

empirically validated for accuracy and efficiency

even though in order to analyse employee attrition

consistency an analysis on special cases (0,0.5,1) of

risk metrics during every phase of framework has

been presented in (Table 3).

Table 3: Special cases of EAI, JS-SAT, EMP-SAT for

employee attrition mitigation.

DE IA DIP EAI

Best 0 0 0 0.163

Average 0.5 0.5 0.5 0.5

Worst 1 1 1 0.837

Employee Attrition Indicator at

Initialization and Recognition phase

OP-OC TR-CG EAI JS-SAT

Best 1 1 0 0.92

Average 0.5 0.5 0.5 0.5

Worst 0 0 1 0.08

Job satisfaction during mitigation phase

APP-

CRR

FLEX-

WLB

EAI EMP-

SAT

Best 1 1 0 0.92

Average 0.5 0.5 0.5 0.5

Worst 0 0 1 0.08

Employee Satisfaction during mitigation phase

JS-SAT EMP-

SAT

EMP-

ATT

Best 0 0 0.92

Average 0.5 0.5 0.5

Worst 1 1 0.08

Employee Attrition during Mitigation

phase

The following stage is to formulate several suggestive

measures based on the analysis carried out in the

previous step. These actions will serve as suggestions

for reducing staff attrition. These recommendations

will help control the risk metrics' values and lessen

employee churn in the IT industry when software

projects are being developed. As a result, the staff

members engaged in the software development

process' risk recognition and mitigation phase have

the following recommendations made for them.

a) Recognize the change in engagement,

performance, absenteeism by taking feedback from

employees. Target the threshold of 20% of

recognized variables. On the basis of last work

documents of company recognized variables will be

updated according to organization environmental

conditions. Value greater than or equal to 25% will

undergo mitigation process, as this will impact on the

percentage of employee attrition.

b) On the basis of value of DE (Somewhat, More),

DIP (Moderate, Substantial), IA (Moderate,

Substantial) in recognition phase based on feedback,

interview, past work document, strategies are

followed according to the fuzzy rules implemented. If

FEMIB 2025 - 7th International Conference on Finance, Economics, Management and IT Business

110

DE, IA, DIP is more than threshold then EAI will be

average and worst.

c) Strategies may be changed throughout the software

development life cycle on the basis of organization

environment feedback.

d) In this study OP-OC, TR-CG, must be changed

during phases of the software life cycle towards 100%

for better Job Satisfaction and better reduced

employee attrition.

e) In this study APP-CRR, FLEX-WLB must be

changed during phases of the software life cycle

towards 100% for better Job Satisfaction and better

reduced employee attrition.

f) JS-SAT and EMP-SAT both are directly

proportional to EMP-ATT (Employee Attrition).

density.

g) Job satisfaction and employee satisfaction level

must be above 25% for better employee attrition.

In the light of above guidelines, the following

recommendations are made to the designer in order to

mitigate employee attrition for smooth functioning of

all phases of software development process.

Continuously monitor the effectiveness of the

implemented strategies and adjust the FIS and rules

as needed to reflect changes in the organization or

industry trends.

4 EMPIRICAL VALIDATION OF

THE MMEA

In order to statistically validate the proposed model

(MMEA), this section of the work calculates the

Pearson's correlation coefficient between the actual

employee attrition values, which are already known,

and the defuzzified (predicted) values using a model

that is used in an IT organization's software

development process to reduce employee attrition.

The researcher contacted reputable and well-

established software development companies in

Noida and Lucknow to confirm or validate the

model's ability to quantify. The researcher then

gathered pertinent data during all phases of software

development life cycle of 16 software projects that

were already implemented and operating (see

appendix). The Table 4 indicate actual data quantified

from above mentioned dataset and predicted data

from proposed research framework.

Table 4: Actual and predicted values.

PROJECTS ACTUAL PREDICTED

1 0.4 0.372

2 0.49 0.473

3 0.4 0.366

4 0.45 0.35

5 0.58 0.5

6 0.39 0.322

7 0.313 0.25

8 0.236 0.2

9 0.45 0.3

10 0.55 0.42

11 0.54 0.42

12 0.7 0.6

13 0.6 0.55

14 0.58 0.52

15 0.45 0.42

16 0.43 0.38

In order to validate the proposed model (MMEA),

EMP-ATT has been computed using the fuzzy

toolbox of MATLAB, for 16 software projects, those

are currently in operation. The related real values and

their predicted values are shown in the (Table 4). The

Pearson's correlation coefficient between anticipated

and actual employee attrition has now been calculated

to verify the model's capacity to be quantified.

Figure 14: SPSS correlation analysis.

The correlation was calculated using IBM SPSS,

and as (Figure 14) illustrates, its value is (0.939). The

correlation value makes it clear that there is a

substantial association between the employee

attrition values that are already known and the

attrition that the MMEA predicts and mitigates. As a

result, it can be said that the suggested model

effectively quantifies staff attrition (Priambodo et al.,

2022).

Fuzzy Based Model for Mitigating Employee Attrition

111

5 RESULTS AND DISCUSSIONS

5.1 Comparison on Employee Attrition

Values

Following section is going to briefly describe and

quantitatively compare some of those studies on the

basis of their relevance with the new Model for

Mitigating Employee Attrition (MMEA) in terms of

their quantified employee attrition values.

(Archita Banerjee, Rahul Kumar Ghosh,

Meghdoot Ghosh, 2017) (Figure 15) contributed to

the West Bengal IT sector and developed a model for

employee retention. The equation that resulted from

the procedure is as follows:

Y= 2.897 - 0.864X1 - 0.305X2 + 0.174X3 +

0.630X4

(1)

Where, Y denote Possibility of staying in the existing

organization, X1 is Uncongenial Organizational

Culture, X2 is Insufficient Compensation follows, X3

is Job Satisfaction, X4 is Sociable Organizational

Practice

Looking at the table values, it is clear that the

MMEA developed in this study predicts and manages

the process of reducing employee attrition in the IT

industry during the software development process

more accurately than the model developed by Archita

Banerjee, Rahul Kumar Ghosh, and Meghdoot Ghosh

(2017) as (shown in Figure 15).

Figure 15: Comparison between proposed model and

existing model.

Satpal, Rajbir Singh and Manju Dhillon (2019)

provided a model for the constructs that exist in the

literature, but only selected dimensions of both

constructs are utilized to generate inferences that

assist companies in identifying factors that influence

attrition intentions. The equation which emerged after

the process was as

Attrition Intentions (C) = 4.884 + 0.215 × H

Factors + 0.201 × Personal Factors + 0.218 ×

Job Related Factors + 0.166 × Organizational

Factors.

(2)

The study attempts to investigate and establish a

relationship between a number of characteristics that

may contribute to retention risks. The study also aims

to draw attention to the shift in tactics used to lower

staff attrition.

Now looking at the table values it can be easily

inferred that the MMEA developed in this research

predict and manage the process of mitigation of

employee attrition in IT industry during software

development process quiet accurately than the model

developed by (Satpal and Dhillon, 2019) as (shown in

Figure 16).

Figure 16: Comparison between proposed model and

existing model.

Deepesh Mamtani and Dr. Bharti Malukani

(2023) suggested a model that focuses on making

precise predictions about employee attrition, needing

a suitable dataset for training and validation reasons.

The implemented machine learning methods are

thoroughly examined, and the results are compiled.

FEMIB 2025 - 7th International Conference on Finance, Economics, Management and IT Business

112

The major goal of this work is to create and apply a

prediction model that can effectively forecast staff

attrition inside a corporation. Proposed model for

employee attrition is expressed in equation which

emerged after the process of logistic regression is

lnactive% = - 3.7*satisfaction_level + 0.20 *

evaluation_score + 0.170 *number_of_years +

0.18

(3)

The study seeks to investigate and establish a link

between numerous elements that may be responsible

for retention risk. Furthermore, the study attempts to

highlight the shift in techniques used and evolving

with the concept of employee engagement to reduce

staff attrition. Looking at the table values, it is clear

that the MMEA developed in this study predicts and

manages the process of reducing employee attrition in

the IT industry during the software development

process more accurately than the model developed by

(Mamtani and Malukani, 2023), as (shown in Figure

17).

Figure 17: Comparison between proposed model and

existing model.

5.2 Measures of Predictive Accuracy

Along with validating a model, guaranteeing its

predicted accuracy is a vital component of any

models development. Any improvement in the

accuracy of employee attrition prediction can have a

major impact on the quality of the software product

under development. The literature shows that the

most popular measures are Magnitude Square Error

(MSE), Mean Magnitude of Relative Error (MMRE),

Balanced MMRE, Mean Absolute Percentage Error

(MAPE), and Prediction at level n (Pred(n)). The

researcher used MATLAB fuzzy toolbox to forecast

and reduce employee attrition of software projects

that are part of the data set by calculating job

satisfaction and employee satisfaction during the

software development process. Table 5 displays the

actual and expected employee attrition values for

each of the 16 projects and predictive accuracy of the

model through these values.

The MMRE number is highly encouraging, falling

significantly below the acceptability criterion of 0.25.

Conte et al recommend MMRE ≤ 0.25 accepted as a

prediction accuracy for prediction model. The

Balanced Mean Magnitude of Relative Error

(BMMRE) and Mean Absolute Percentage Error

(MAPE) are the next important accuracy metrics to

calculate after the MMRE as shown in Figure 18. It is

evident from the figures of the several accuracy

metrics that the Model for Mitigating Employee

Attritions has a reasonably accurate prediction ability.

Consequently, the model may be applied to precisely

predict, quantify, and reduce employee attrition

across the software development process and life

cycle. Given that the errors are less than half the

difference between two output outcomes, the model's

validation showed satisfactory validity.

Figure 18:

Measures of Predictive Accuracy for MMEA

Model.

ACTUAL PREDICTED ERROR

ABS OF ERROR Square of error MRE

BMRE %ERROR

1 0.4 0.372 0.0280 0.0280 0.0008 0.0700 0.0753 7.0000

2 0.49 0.473 0.0170 0.0170 0.0003 0.0347 0.0359 3.4694

3 0.4 0.366 0.0340 0.0340 0.0012 0.0850 0.0929 8.5000

4 0.45 0.35 0.1000 0.1000 0.0100 0.2222 0.2857 22.2222

5 0.58 0.5 0.0800 0.0800 0.0064 0.1379 0.1600 13.7931

6 0.39 0.322 0.0680 0.0680 0.0046 0.1744 0.2112 17.4359

7 0.313 0.25 0.0630 0.0630 0.0040 0.2013 0.2520 20.1278

8 0.236 0.2 0.0360 0.0360 0.0013 0.1525 0.1800 15.2542

9 0.45 0.3 0.1500 0.1500 0.0225 0.3333 0.5000 33.3333

10 0.55 0.42 0.1300 0.1300 0.0169 0.2364 0.3095 23.6364

11 0.54 0.42 0.1200 0.1200 0.0144 0.2222 0.2857 22.2222

12 0.7 0.6 0.1000 0.1000 0.0100 0.1429 0.1667 14.2857

13 0.6 0.55 0.0500 0.0500 0.0025 0.0833 0.0909 8.3333

14 0.58 0.52 0.0600 0.0600 0.0036 0.1034 0.1154 10.3448

15 0.45 0.42 0.0300 0.0300 0.0009 0.0667 0.0714 6.6667

16 0.43 0.38 0.0500 0.0500 0.0025 0.1163 0.1316 11.6279

0.1018 0.1489

2.9642 238.2530

MSE(Mean Square

Error)

0.006363625

RMSE(Root Mean

Square Error)

0.079772332

MMRE/MPE(Mean

Magnitude of

relative error/Mean

percentage error)

0.009306758

BMMRE(Balanced

MMRE)

0.185262887

MAPE(Mean

Absolute percentage

error)

14.89081323

PREDICTION AT

LEVEL 0.25

Pred(0.25)

93.75 93.75% of predicted EMP-ATT value by EAMM have MRE's less than or equal to 0.25

PROJECTS

Fuzzy Based Model for Mitigating Employee Attrition

113

Looking at the values of various accuracy measures,

it is evident that prediction ability of the Employee

Attrition Mitigation Model is quiet accurate.

Therefore, it can be concluded that the model can be

used to accurately predict, track and mitigate

employee attrition during software development life

cycle during software development process. The

errors validated by the model exhibited satisfactory

validity, as they are less than half the distance

between two output results (Nikmanesh,2023).

5.3 Comparison on Correlation

Coefficient

The researcher has computed the Pearson’s

Correlation Coefficient between the predicted values

of employee attrition (through the proposed model;

Archita Banerjee, Rahul Kumar Ghosh, Meghdoot

Ghosh (2017); Satpal, Rajbir Singh and Manju

Dhillon (2019); Deepesh Mamtani, Dr. Bharti

Malukani (2023)) and the actual values of the

employee attrition. Looking at the values of the

following table it can be easily noticed that the

proposed model in this research has a very High

Positive Correlation, While the research work done

by Archita Banerjee, Rahul Kumar Ghosh, Meghdoot

Ghosh(2017) has High Negative Correlation, and the

work done in the same area by Satpal, Rajbir Singh

and Manju Dhillon (2019) has a Moderate Negative

Correlation, and research study by Deepesh Mamtani,

Dr. Bharti Malukani (2023) produces Moderate

Positive Correlation (Figure 19 –Figure 22).

Proposed Model Pearson’s Correlation Coefficient Measure

Figure 19: Correlation analysis of proposed model.

Archita Banerjee, Rahul Kumar Ghosh, Meghdoot Ghosh

(2017) Pearson’s Correlation Coefficient measure (Banerjee et

al., 2017)

Figure 20: Correlation analysis of model developed in

2017.

Satpal, Rajbir Singh and Manju Dhillon (2019) Pearson’s

Correlation Coefficient measure (Satpal and Dhillon, 2019)

Figure 21: Correlation analysis of model developed in

2019.

Deepesh Mamtani, Dr. Bharti Malukani (2023) Pearson’s

Correlation Coefficient (Mamtani and Malukani, 2023)

Figure 22: Correlation analysis of model developed in

2023.

FEMIB 2025 - 7th International Conference on Finance, Economics, Management and IT Business

114

Table 5: Correlation levels of proposed model with other

models.

No.

Employee Attrition

Model

Pearson's

Correlation

Coefficient

Correlation

level

1 Archita Banerjee, Rahul

Kumar Ghosh, Meghdoot

Ghosh(2017) (Figure 20)

-920 High

Negative

2 Satpal, Rajbir Singh and

Manju Dhillon (2019)

(Figure 21)

-510 Moderate

negative

3 Deepesh Mamtani, Dr.

Bharti Malukani (2023)

(Figure 22)

0.360 Low positive

4 Proposed Model(MMEA)

(Figure 19)

0.939 High Positive

Therefore, it can be concluding that the model

(MMEA) of this research is better than the three

existing models, on the basis of quantitative values

(Table 5).

6 CONCLUSION AND FUTURE

EXTENSION

This study could serve as the basis for future research

for risk mitigation in software organizations. The

framework is quite prescriptive in nature, and will

definitely facilitate industry professionals and

researchers to recognize and reduce employee

attrition during software development life cycle

process of software development in IT industry.

Consideration of the employee attrition indicator

along with employee attrition effected by other risk

factors on the basis of its value is an edge over other

studies those are based on only prediction and

considering employee data because ignoring or

overlooking indicator factors and only concentrating

on making the risk metrics will not seem good

enough.

The MMEA model developed as per the

guidelines of the proposed framework in analysis and

finalization phase in section 4 may help software

professionals to take appropriate corrective measures

right from starting phase and continuing towards

other phases on the basis of immune theoretical

concept of primary measures and secondary measures

to help designers as well as developers to predict and

reduce employee attrition during software

development process in the software development life

cycle with an improved efficiency and quality level.

The research has utilized the strength of fuzzy

inference process in building model. The assessment

and amendment of the framework further strengthens

it practicality as well as viability by keeping the doors

of improvement open for any of the earlier phases. In

most of the cases, developed models only provide

quantitative values but neither provides suggestions

on how to make improvement, nor the precautions on

how to avoid abnormalities. Therefore, to fill this gap

research has provided the suggestive measures and

recommendations based on the results and contextual

interpretations.

Apart from the above, reassessment of previously

developed or underdevelopment employee attrition

models could be done as per the guidance proposed

as well as recommendation in this study (Gupta,

2022), (Wardhani and Lhaksmana, 2022),

(Udechukwu and Mujtaba, 2007). Beside this, as far

as further research is concern, the model may open

fresh avenues for the researchers, doing research on

employee attrition estimation as well as dealing with

strategies to overcome employee attrition. Validating

and testing the suggested risk mitigation procedure

against other common risk factors occur during

Software development life cycle in an actual setting

is one way to conduct additional.

ACKNOWLEDGEMENTS

I would like to express my special thanks of gratitude

to Dr. Wajahat Abbas Rizvi as well as Dr. Vinodani

Katiyar who helped me to do this wonderful research

work, which also helped me in doing a lot of Research

and I came to know about so many new things I am

really thankful to them.

Secondly I would also like to thank my family and

friends who helped me a lot in finalizing this study

within the limited time frame.

REFERENCES

H Shihadeh, T. (2014). Risk Factors in Software

Development Phases. European Scientific Journal, 10,

213–231.

Raza, A. (2022). Predicting Employee Attrition Using

Machine Learning Approaches. Applied Sciences,

12(13).

Banerjee, A. (2017). A Study on the Factors Influencing the

Rate of Attrition in IT Sector Based on Indian Scenario.

Pacific Business Review International, 9, 10–19.

Satpal, S., & Dhillon, M. (2019). Impact of Factors

Affecting Employee Retention on Attrition Intentions:

In Indian IT Sector. International Journal on Emerging

Technologies, 10(4), 273–282.

Mamtani, D., & Malukani, B. (2023). Predictive Model for

Employee Attrition Risk Assessment. European

Economic Letters (EEL), 13(1s), 358–364.

Fuzzy Based Model for Mitigating Employee Attrition

115

Yadav, H. B., & Yadav, D. K. (2015). A Fuzzy Logic Based

Approach for Phase- Wise Soft-Ware Defects

Prediction Using Software Metrics. Information and

Software Technology, 63, 44–57.

Nikmanesh, M. (2023). Employee Productivity Assessment

Using Fuzzy Inference System. Employee Productivity

Assessment Using Fuzzy Inference System. Information,

14.

Ahmed, I. (2013). Employee Performance Evaluation: A

Fuzzy Approach. International Journal of Productivity

and Performance Management, 62, 718–734.

Sharma, M. K. (2022). Employee Retention and Attrition

Analysis: A Novel Approach On Attrition Prediction

Using Fuzzy Inference and Ensemble Machine

Learning. Webology, (2).

Urs, N. S. (2014). Exploring the Dynamics of Job

Satisfaction and Employee Engagement in IT/ITes

Industries. Journal of Applied Management and

Investments, 3(1).

Kannan, M., & Vivekanandan, K. (2012). Study on Attrition

among New Entrants in Soft-Ware Testing

Professionals.

Demirel, Z., & Çubukçu, C. (2021). Measurement of

Employees on Human Resources with Fuzzy Logic.

EMAJ: Emerging Markets Journal, 11, 1–7.

Gupta, Santosh & Bhatia, Dr. Nitesh. (2023). A Decade of

Trend in the Employee Turnover Intention Study in

India: A Systematic Review and Recommendation.

FIIB Business Review. 10.1177/23197145231158907.

Bukohwo, E. M. (2015). Risk Model for Software

Development Personnel. Proceedings of the

International Multi Conference of Engineers and

Computer Scientists, 1.

Jiang, R. (2015). A Novel Risk Metric for Staff Turnover in

a Software Project Based on Information Entropy.

Software Project Based on Information Entropy:

Entropy, 17, 2834–2852.

Díaz, G. (2023). Analysing Employee Attrition Using

Explainable AI for Strategic HR Decision-Making.

Mathematics, 11.

Gupta, S. K. (2022). A Review of Employee Attrition

Models and Their Role in Evolution of Attrition

Literature. The Indian Journal of Labour Economics,

65(1), 185–214.

Govindaraju, N. (2018). Addressing Employee Turnover

Problem: A Review of Employee Turnover Core

Models. International Journal of Innovative Science

and Research Technology, 11, 516–527.

Srivastava, P. R., & Eachempati, P. (2021). Intelligent

Employee Retention System for Attrition Rate Analysis

and Churn Prediction: An Ensemble Machine Learning

and Multi-Criteria Decision-Making Approach.

Journal of Global Information Management (JGIM),

29(6), 1–29.

El-Rayes, N. (2020). Predicting Employee Attrition Using

Tree-Based Models. International Journal of

Organizational Analysis, 28(6), 1273–1291.

Priambodo, Bagus & Jumaryadi, Yuwan & Rahayu,

Sarwati & Ani, Nur & Ratnasari, Anita & Salamah,

Umniy & Putra, Zico & Otong, Muhamad. (2022).

Predicting Employee Turnover in IT Industries using

Correlation and Chi-Square Visualization. International

Journal of Advanced Computer Science and

Applications. 13. 10.14569/IJACSA.2022.0131210.

Wardhani, F. H., & Lhaksmana, K. M. (2022). Predicting

Employee Attrition Using Logistic Regression with

Feature Selection. Sinkron: Jurnal Dan Penelitian

Teknik Informatika, 6, 2214–2222.

Priya, V. K., & Harasudha, H. H. (2017). A Study on

Employee Attrition with Reference to Lanson Toyoya,

Chennai. Chennai. Man in India, 97, 115–124.

Udechukwu, I. I., & Mujtaba, B. G. (2007). Determining

the Probability That an Employee Will Stay or Leave

the Organization: A Mathematical and Theoretical

Model for Organizations. Human Resource

Development Review, 6(2), 164–184.

Eyupoglu, S. Z. (2017). Job Satisfaction: An Evaluation

Using a Fuzzy Approach. Procedia Computer Science,

120, 691–698.

Thirupathy, A., & Dhayalan, C. (2016). Employee

Retention and Turnover Using Motivational Variables

at India. Int. J. Res. Granthaalayah, 4(8), 1–9.

Purohit, M. (2016). A Study On-Employee Attrition in IT

Sector with Special Emphasis on Wipro and Infosys.

IOSR Journal of Business and Management, 18(4), 47–

51.

Singh, R., & Satpal, M. (2018). Factors Affecting

Employee Retention in Indian IT Sector.

Abhyankar, A. (2021).: Indian IT Industry: High Attrition

Rates and Employee Retention Strategies. Webology,

Vol 18, No.1, 2023-2029

Teotia, S. (2023). Factors Affecting Employee Retention:

An Analysis of It Sector in Delhi-NCR (pp. 94–98).

Noble Science Press.

Harikumar Pallathadka, V. H. (2022). Attrition in Software

Companies: Reason and Measures. Materials Today:

Proceedings, 51, 528–531.

Saraf, V., & Peshave, D. M. A. (2020). An Analysis on

Employee-Attrition in IT Industry. Mukt Shabd

Journal.

Rusi, X., & Viollet, B. (2023). Factors Affecting Employee

Attrition: A Systematic Literature Review. Economic

Restructuring for Sustainable Development.

Singh, K., & Singh, R. (2019). A Study on Employee

Attrition: Effects and Causes. International Journal of

Research in Engineering, 2(8), 2581–5792.

David, S. (2015). Attrition in IT Sector. International

Journal of Core Engineering and Management

(IJCEM), 2(1), 74–92.

Kermani, A. G. (2021). Human Resource Risk

Management Framework and Factors Influencing IT.

Propósitos y Representac-Iones, 9.

Popescu, S. (2020). A Structured Framework for

Identifying Risks Sources Related to Human Resources

in a 4.0 Working Environment Perspective. 511–527.

Chauhan, V. S., & Patel, D. (2013). Employee Attrition: A

Factorial Study of IT Industry. Journal of Strategic

Human Resource Management, 2(1).

FEMIB 2025 - 7th International Conference on Finance, Economics, Management and IT Business

116

Saher, N. (2015). The Impact of Employees Attrition at the

Productivity of a Software. International Journal of

Natural and Engineering Science, 9(3), 23–37.

Rizvi, S. W. A., Singh, V. K., & Khan, R. A. (2017). Early

stage software reliability modeling using requirements

and object-oriented design metrics: fuzzy logic

perspective. International journal of computer

applications, 162(2), 44-59.

Patil, R. S. (2011). Human Resource Challenges &

Practices in IT Industry. In Proceedings of the 5th

National Conference (pp. 10–11).

Prasad, A., & Kamalanabhan, P. (2010). Human Resource

Excellence in Software Industry in India: An

Exploratory Study. International Journal of Logistics

Economics and Globalization, 2(4), 316–330.

Flouris, T., & Yilmaz, A. K. (2010). The Risk Management

Framework to Strategic Human Re-Source

Management. International Research Journal of

Finance and Economics, 36(1), 25–45.

Alqahtani, Haya & Almagrabi, Hana & Alharbi, Amal.

(2024). Employee Attrition Prediction using Machine

Learning Models: A Review Paper. International

Journal of Artificial Intelligence & Applications. 15.

23-49. 10.5121/ijaia.2024.15202.

George, S. (n.d.). Predicting Employee Attrition Using

Machine Learning Algorithms. 2022 4th International

Conference on Advances in Computing,

Communication Control and Networking (ICAC3N).

Almohammadi, K. (2024). Adaptive Type-1 Fuzzy Logic-

Based System for Predicting Em-Ployee Attrition.

International Journal of Intelligent Systems and

Applications in Engi-Neering, 12(21s).

Shete, M. (2021). Prediction of Employee Attrition Using

Machine Learning Approach. Inter-National Journal for

Research in Applied Science and Engineering

Technology, (9).

Bormah, A., Taylan, O. Productivity Forecasting of

Employees Performance Using Machine Learning and

Adaptive Neuro-Fuzzy Inference System. International

Journal of Science, Engineering and Technology. 11.

Arslankaya, Seher. (2023). Comparison of performances of

fuzzy logic and adaptive neuro-fuzzy inference system

(ANFIS) for estimating employee labor loss. Journal of

Engineering Research. 11. 100107.

10.1016/j.jer.2023.100107.

Hasib, N., Rizvi, S. W. A., & Katiyar, V. (2023). Artificial

Immune System: A Systematic Literature Review.

Journal of Theoretical and Applied Information

Technology, 101(4), 1469–1486.

Hasib, Nida, Rizvi, S. W. A., & Katiyar, V. (2024).

Conceptual framework for risk mitigation and

monitoring in software organizations based on artificial

immune system. In Lecture Notes in Networks and

Systems. Lecture Notes in Networks and Systems (pp.

25–37). doi:10.1007/978-981-97-7423-4_3

Hasib, N., Rizvi, S. W. A., & Katiyar, V. (2023). Biological

immune system based risk mitigation monitoring

system: An analogy. International Conference on

Artificial Intelligence, Blockchain, Computing and

Security ICABCS 2023, Volume 1 (pp. 760-767). CRC

Press. DOI.10.1201/9781003393580-113.

Hasib, N., Rizvi, S. W. A., & Katiyar, V. (2023). Risk

Mitigation and Monitoring Challenges in Software

Organizations: A Morphological Analysis.

International Journal on Recent and Innovation Trends

in Computing and Communication, 11(8), 172–185.

https://doi.org/10.17762/ijritcc.v11i8.7943.

Hasib, N., Rizvi, S. W. A., & Katiyar, V. (2025). A Review

of Risk Factors Affecting Employee Attrition in IT

Sector. 5th International Conference on Paradigms of

Communication, Computing and Data Analytics

(PCCDA 2025), 18 Jan – 20 Jan, Pt. Lalit Mohan

Sharma Campus, Rishikesh, Sri Dev Suman

Uttarakhand University, Uttarakhand, India. (Accepted

and preseted)(Unpublished).

APPENDIX

The questionnaires used for conducting the

qualitative exploratory study and quantitative

descriptive study are listed in this section.

Appendix A1 Qualitative Exploratory Study(In-depth

Interviews/Work Document)

Appendix A2

Appendix B1

Appendix B2

 Part I

 Part II

Quantitative Descriptive Survey -Risk Factor

Ranking

 Pre-Intervention Questionnaire for

Validation of Employee Attrition Mitigation

Framework in Software Development

Projects during SDLC

 Post Intervention Questionnaire for

Validation of Employee Attrition Mitigation

Framework in Software Development

Projects during SDLC

1) Outcome Assessment of the Framework

in Terms of Predicted Value

2) Qualitative Reviews

Fuzzy Based Model for Mitigating Employee Attrition

117