Early Detection of Employee Turnover Risks Using Machine

Learning Approaches

Lakshmi Satwika Neelisetty, Naga Preethika Reddy Bonthu and Amar Jukuntla

Department of ACSE, VFSTR deemed to be University, Guntur, Andhra Pradesh, India

Keywords: Employee Attrition, Turnover Prediction, Machine Learning, Workforce Retention, Predictive Analytics,

Human Resource Analytics, Attrition Risk Assessment.

Abstract: Employee turnover poses a significant challenge for organizations, resulting in productivity losses and

increased costs associated with recruitment, training, and knowledge transfer. Predicting attrition in advance

allows organizations to implement proactive retention strategies, thereby improving workforce stability. This

study proposes a machine learning-based predictive model to identify employees at risk of leaving by

analyzing key factors such as demographic attributes, job roles, performance metrics, and organizational

influences. By leveraging advanced data-driven techniques, the model estimates the likelihood of attrition,

providing actionable insights for HR decision-making. The proposed approach aims to enhance employee

retention efforts by enabling organizations to address underlying factors contributing to turnover, ultimately

fostering a more engaged and stable workforce.

1 INTRODUCTION

Employee attrition is a critical challenge for

organizations across various industries, leading to

significant financial and operational impacts.

Attrition, whether voluntary or involuntary, results in

increased recruitment and training costs, loss of

institutional knowledge, and reduced productivity.

Moreover, frequent turnover can disrupt workflow,

affect team morale, and hinder long-term

organizational growth. In an increasingly competitive

job market, organizations must proactively identify

key factors influencing employee departures to

enhance retention strategies and maintain workforce

stability.

Predicting employee attrition in advance can

provide valuable insights into turnover patterns,

enabling companies to implement targeted

interventions. Machine learning techniques offer a

data-driven approach to analyzing attrition by

considering various employee attributes such as

demographic information, job roles, salary levels,

tenure, performance metrics, and work-life balance.

Prior research has utilized models like logistic

regression, Support Vector Machines (SVM), and

Random Forest, achieving prediction accuracy

ranging from 82% to 87%. While these approaches

have demonstrated effectiveness, improvements in

feature selection, model tuning, and advanced deep

learning methods can further enhance predictive

accuracy.

This study aims to develop an optimized

machine learning model for employee attrition

prediction by evaluating multiple algorithms and

feature sets. The proposed model will not only

identify employees at risk of leaving but also provide

insights into the most influential factors driving

attrition. The findings of this research will support

human resource departments in making informed

decisions, optimizing retention strategies, and

improving overall workforce management.

The remainder of this paper is structured as

follows. The next section provides a literature review,

discussing previous works and methodologies

applied in employee attrition prediction. The

methodology section details the dataset,

preprocessing techniques, and machine learning

models used in this study. The results and

performance analysis section presents experimental

findings, including accuracy, precision, recall, and

F1-score comparisons. Finally, the conclusion

highlights key insights, practical implications, and

potential future research directions.

Neelisetty, L. S., Bonthu, N. P. R. and Jukuntla, A.

Early Detection of Employee Turnover Risks Using Machine Learning Approaches.

DOI: 10.5220/0013893000004919

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

In Proceedings of the 1st International Conference on Research and Development in Information, Communication, and Computing Technologies (ICRDICCT‘25 2025) - Volume 3, pages

119-128

ISBN: 978-989-758-777-1

119

2 LITERATURE SURVEY

In Meraliyev et al. 2023 paper authors introduced a

model "Attrition Rate Measuring In Human Resource

Analytics Using Machine Learning" In this study, the

authors have done work towards predicting employee

attrition with machine learning techniques. The

dataset started with 21 employee-related features,

retaining only those that were found statistically

significant with respect to attrition status. Categorial

variances like gender and job type were changed into

numeric forms to make the dataset amenable to

machine learning. Dummy variables were created for

categorical features with different string values to turn

the dataset into 269 columns. The missing values in

the columns age and experience were treated by

replacing them with the mean, thereby making the

model efficient. Multiple classification models were

applied to ascertain attrition. All models were

evaluated using a confusion matrix that calculates

accuracy, precision, and recall. Out of the four models

analyzed, logistic regression was found to be the most

accurate at 75%, making it the most viable model in

the analysis. Multinomial Naive Bayes, a very well-

known algorithm usually employed for text

classification, did not work well because of the type

of data. The K-NN was slightly more effective

because it determines the class of the data point based

on the proximity of some previously classified points.

Gaussian Naive Bayes was also considered, but

performance details are largely omitted. The study

indeed emphasizes the importance of data

preprocessing and feature selection in optimizing the

model performance, especially the efficiency of

Logistic Regression in predicting employee attrition.

In Mitravinda, K. M., and Sakshi Shetty paper

authors introduced a model "Employee Attrition:

Prediction, Analysis of Contributory Factors and

Recommendations for Employee Retention" in

(2022). A recommendation system was built for

providing the employer with recommendations of

how attrition can be prevented for a newly input

record of an employee using Logistic Regression,

XGBoost, Adaboost, KNN acheiving the accuracies

of 87.075 %, 87.074%, 85.714%, 84.353%. The best

performing model XGBoost was used to obtain the

SHAP index for all the instances in the dataset.

Figure

1 shows the machine learning workflow for attrition

prediction.

In Yadav et al. 2018 paper authors proposed "Early

Prediction of Employee Attrition using Data Mining

Techniques" One such recent paper compared the

classification of ten departmental categories into two

categories Technical and Non-Technical by brute-

force approach and One-Hot Encoding to prevent bias

in the machine learning model. Various classifiers

were used, among which Decision Tree achieved the

maximum accuracy (99.51%), followed by Random

Forest (99.05%) and AdaBoost (95.99%). Additional

model optimizations to AdaBoost and Random Forest

provided incremental results, among which AdaBoost

achieved the maximum number of instances correctly

classified (1441). The paper brings to the forefront the

necessity of optimal feature encoding to enable

improved classification performance and balanced

learning between categories.

Figure 1: Machine Learning Workflow for Attrition

Prediction.

In (Maharana et al. 2022) papers the authors

proposed a model" Automated Early Prediction of

Employee Attrition in Industry Using Machine

Learning Algorithms" This project seeks to predict

the employee attendance utilizing classification

techniques, such as Decision Tree, Logistic

Regression, and Random Forest. Using Google

Colaboratory, it involves understanding the

problems, collecting employee data, data

preprocessing (cleaning, normalization, and

transformation), and exploratory data analysis. The

dataset is split into a training set and a test set to build,

compare, and rate with performance metrics several

classification models. Some of the key preprocessing

included removal of irrelevant variables, provision of

ICRDICCT‘25 2025 - INTERNATIONAL CONFERENCE ON RESEARCH AND DEVELOPMENT IN INFORMATION,

COMMUNICATION, AND COMPUTING TECHNOLOGIES

120

categorical data, and consistency of the data. The

correlation analysis shows very strong relationships

among the Job Level, Total Working Years, and

Monthly Income relating to the insightful objectives

identified for attrition prediction achieving the

accuracy of 86%, 88%, 89% for Random forest,

Decision tree, and Logistic Regression respectively

(table 1).

Table 1: Comparison of Machine Learning Models for

Employee Attrition Prediction.

Machine Learning Model Accuracy (%)

Logistic Regression 75 - 88

Random Forest 83 - 99.05

XGBoost (Best Model) 87.07

Decision Tree 86 - 99.51

K-NN 80 - 85.71

Naive Bayes 72

In (Polisetti et al.2024) paper the authors

proposed a model "Stacking Models for Employee

Attrition Prediction: Leveraging Logistic Regression

and Random Forest" Attrition analysis is a systematic

exploration of employee turnover through data-

centric techniques. This method starts from the data

sources of HR reports or employee surveys and exit

interviews, which entails preprocessing the data for

quality. EDA is used to determine any patterns or

correlations that influence attrition. After this, feature

selection is applied in order to discover significant

variables for predictive modeling; some of the

techniques would include logistic regression,

decision trees, and neural networks. Model

evaluation is done with accuracy, precision, and

recall metrics for augmentation of robustness through

cross-validation. From the models evaluated,

combining logistic regression with random forest

proved to be the most accurate (90%), followed by

boosting (88.5%), random forest (83%), logistic

regression (81\%), KNN (80%), and naive Bayes

(72%). The consequence of using all these models, in

essence, is that they would help organizations to put

in place targeted retention policies that would slow

turnover down alongside increasing employee

satisfaction. Table 2 gives the comparison of related

work on employee attrition prediction.

Table 2: Comparison of Related Work on Employee Attrition Prediction.

Reference

No.

Method Used Advantage Drawback

Zheng et al.,

2022

Logistic Regression, Random

Forest

High accuracy, simple

implementation

Limited feature

interaction analysis

Subasri et al.,

2023

Random Forest, Logistic

ression

Good performance with tabular

data

Poor performance on

small datasets

Arora et al.,

2017

SVM, XGBoost, Decision Trees High recall with SVM

High computational

cost for complex

models

Occhipinti et

al., 2022

Hybrid Models (Ensemble

Techni

ues

)

Improved robustness

Increased complexity

trainin

time

Yildiz et al.,

2017

Comparative Study of ML

Models

Comprehensive model evaluation

Lack of deep learning

models consideration

Minaee et al.,

2020

Deep Learning Models High predictive accuracy

Requires large datasets

extensive tuning

The paper is organized as follows: Section 2

presents the literature survey, providing a

comprehensive review of existing research related to

employee attrition prediction. Section 3 describes the

methodology employed, detailing the data collection,

preprocessing techniques, and feature selection

methods utilized. Section 4 outlines the design of the

proposed application for practical implementation.

Section 5 discusses various classification models

used for employee attrition prediction, including their

underlying principles and implementation. Section 6

focuses on the performance metrics applied to

evaluate model accuracy and robustness. Section 7

provides a detailed comparison of model

performance based on the chosen metrics. Finally, the

conclusion summarizes the findings, highlights

limitations, and suggests directions for future

research.

3 METHODOLOGY

An advanced machine learning methodology for

employee attrition forecasting would allow itself to

rely on correctly formulated predictive models. The

Early Detection of Employee Turnover Risks Using Machine Learning Approaches

121

analysis starts after the requisite data preparation

where the dataset is loaded, and every column name

is transformed into lower case in uniformity. After

that, we should explore if there resides a "target" field

in binary format with "Yes" representing 1 (for left)

and "No" representing 0 (for stayed). The relevant

features selected for analysis could be either

numerical (years at the company, salary) or

categorical (job role, department). All categorical

variables will be encoded by one-hot encoding or

other relevant methodologies that allow their use in

machine learning models. Afterward, handling of

missing values for some of the variables will be

attempted, possibly data scaling or normalization as

another round of preparation.

Figure 2 shows the

Employee Attrition Prediction.

Figure 2: Employee Attrition Prediction.

Feature selection techniques will then be

performed to find the most important predictors of

attrition, distinguishing between training and

validation components for the prediction. Four

models were varied: Logit, support vector machine

(SVM), Random Forest, and XGBoost. Logistic

Regression represents both a baseline and the point

for the conclusion about how significant each feature

is. Besides, SVM is good from its capability to deal

with high dimensionality situations. It is also

dependent on a tree structure like Random Forest to

identify non-linear relationships and feature

importance. The XGBoost comes in bigger to fight in

due to its reliability towards handling big data within

a limited time frame for analysis and conclusions. For

hyperparameter tuning, grid search or random search

in varied settings across each model are utilized for a

performance boost.

Figure 3: Proposed Architecture for Employee Attrition

Prediction.

Cross-validation assures that the models continue

to work in a generalized fashion on the unseen data.

The metrics adopted for performance include

accuracy, precision, recall, F1 score, and ROC-AUC,

which contribute to a more convenient comparison.

Figure 3 illustrates the Proposed Architecture for

Employee Attrition Prediction.

3.1 Algorithm 1: Machine Learning

Model for Employee Attrition

Section 1.01 Input

• Dataset D with features X and labels Y

Section 1.02 Output

• Optimized Model M

Section 1.03 Step 1: Problem Articulation

• Identify the problem statement, objectives, and

key requirements.

Section 1.04 Step 2: Data Collection

• Collect empirical data from reliable sources to

create a comprehensive dataset.

Section 1.05 Step 3: Data Preprocessing

• Noise reduction to enhance data quality.

• Handling missing values using imputation

techniques.

• Feature selection for retaining relevant

attributes.

• ata normalization: see Equation (1)

𝐗′ = (𝐗 − 𝛍) / 𝛔 (1)

Section 1.06 Step 4: Model Implementation

• Train models using various machine learning

techniques:

• For each model M in {Random Forest, SVM,

XGBoost, Logistic Regression, MLP}:

• - Train model on training data D_train.

• - Compute loss function: see Equation (2)

• - Update model parameters: see Equation (3)

ICRDICCT‘25 2025 - INTERNATIONAL CONFERENCE ON RESEARCH AND DEVELOPMENT IN INFORMATION,

COMMUNICATION, AND COMPUTING TECHNOLOGIES

122

𝐋(𝛉) = 𝚺 𝐥(𝐲ᵢ, ŷᵢ)  𝚺 𝛀(𝐟ₜ) (2)

𝐖⁽ˡ⁾ = 𝐖⁽ˡ⁾ − 𝛈 𝛛𝐋/𝛛𝐖⁽ˡ⁾ (3)

Section 1.07 Step 5: Model Testing and Pattern

Detection

• Evaluate model on test dataset D_test.

Section 1.08 Step 6: Performance Analysis and

Optimization

• Compute evaluation metrics:

• Accuracy: see Equation (4)

• Precision: see Equation (5)

• Recall: see Equation (6)

• F1 Score: see Equation (7)

Accuracy =





(4)

Precision =





(5)

Recall =





(6)

F1 = 2

PrecisionRecall

PrecisionRecall

(7)

Section 1.09 Step 7: Model Optimization

• Compare models and update parameters to

maximize accuracy.

Section 1.10 Return

• Optimized Model M

3.2 Application Design

The proposed application integrates various machine

learning models to enable real-time classification and

prediction of employee attrition. The user interface,

shown in Figure 4, allows users to upload datasets in

common formats such as CSV and Excel, preprocess

data, and select appropriate classification models.

Users can fine-tune model parameters and initiate the

training and testing processes, making the application

adaptable to different datasets and prediction

requirements.

Figure 4 shows the user interface for

data upload, model selection, and execution.

To enhance interpretability, the application

provides various visualization tools. These include

confusion matrices, ROC curves, and feature

importance charts, allowing users to evaluate model

performance effectively. The application also

supports exporting analysis results, making it a

practical tool for real-world employee attrition

prediction scenarios. The combination of model

selection, performance comparison, and visualization

features ensures comprehensive insights into model

effectiveness and usability.

Figure 4: User Interface for Data Upload, Model Selection,

and Execution.

3.3 Classification Models for Employee

Attrition Prediction

Employee attrition prediction relies on classification

models to categorize employees into those likely to

leave and those likely to stay. Classification models

are essential in predictive analytics as they leverage

historical employee data to recognize patterns and

identify key factors contributing to attrition. Various

machine learning techniques, such as Logistic

Regression, Decision Trees, Random Forest, Support

Vector Machines (SVM), and Gradient Boosting

methods, have been employed to improve the

accuracy of predictions.

In this section, an overview of these classification

techniques is presented, discussing their

methodologies, advantages, and limitations

concerning employee attrition analysis. The

performance of these models is evaluated using

metrics such as accuracy, precision, recall, and F1-

score. Additionally, ensemble models, which

combine multiple classifiers to enhance predictive

accuracy, are explored.

Understanding the strengths and weaknesses of

different classification techniques allows

organizations to choose the most effective model for

workforce retention strategies. The subsequent

subsections delve into individual models, their

working principles, and their relevance in predicting

employee turnover.

Early Detection of Employee Turnover Risks Using Machine Learning Approaches

123

3.3.1 Random Forest

Random Forest Classifier (RFC) is widely used for

employee attrition prediction due to its robustness

and ability to handle complex datasets. It constructs

multiple decision trees during training and

determines the final classification through majority

voting, reducing the risk of overfitting. By analyzing

various employee attributes such as tenure, salary,

job satisfaction, and performance metrics, RFC

effectively identifies employees at risk of leaving.

The model achieves an accuracy of 88%, making it a

reliable choice for predicting attrition trends and

aiding organizations in workforce retention strategies

(figure 5).

𝑔

(

𝑥

)

=𝑓



(

𝑥

)

𝑓



(

𝑥

)

𝑓



(

𝑥

)

 . . . (8)

Figure 5: Random Forest Classifier.

3.3.2 Support Vector Machine (SVM)

Support Vector Machines are widely employed for

employee attrition prediction due to their ability to

classify employees based on key attributes such as

job satisfaction, salary, and tenure. SVM works by

finding an optimal hyperplane that maximizes the

margin between employees likely to leave and those

who will stay. To enhance performance,

normalization and scaling techniques are applied to

handle complex data distributions. With these

optimizations, SVM achieves an accuracy of 87%,

making it a powerful tool for identifying attrition

risks and aiding HR strategies for employee retention

(figure 6).

Objective Function =



margin

λ∑penalty (9)

Figure 6: SVM Classifier.

3.3.3 XGBoost

XGBoost, a powerful gradient-boosting algorithm, is

widely used for employee attrition prediction due to

its ability to handle complex data patterns. It

efficiently models relationships between employee

attributes such as job role, performance metrics, and

work-life balance to predict the likelihood of

attrition. By leveraging decision trees in an ensemble

learning framework, XGBoost identifies key factors

influencing employee turnover. When optimized with

feature selection and hyperparameter tuning, it

achieves an accuracy of 87%, making it a reliable tool

for workforce retention strategies (figure 7).

ℒ

(

)

∑

𝑙

(

𝑦



,𝑦





)







∑

(

𝑓



)





(10)

𝑦



(



)



=𝑦



(



)



𝑓



(

𝑥



)

(11)

𝑌



∑

𝑓



(

𝑋

)





(12)

Figure 7: XGBoost Classifier.

ICRDICCT‘25 2025 - INTERNATIONAL CONFERENCE ON RESEARCH AND DEVELOPMENT IN INFORMATION,

COMMUNICATION, AND COMPUTING TECHNOLOGIES

124

4 PERFORMANCE METRICS

FOR EMPLOYEE ATTRITION

PREDICTION

To assess the effectiveness of clustering techniques in

employee attrition prediction, various performance

metrics are utilized:

Silhouette Score: Measures how well-separated the

clusters are, with higher values indicating better-

defined clusters.

Davies-Bouldin Index: Evaluates cluster

compactness and separation, where lower values

indicate better clustering.

Adjusted Rand Index (ARI): Compares clustering

results with ground truth labels to measure accuracy.

Normalized Mutual Information (NMI):

Quantifies the shared information between predicted

clusters and actual labels, ensuring meaningful

segmentation.

Table 3: Performance Metrics for Clustering in Employee

Attrition Prediction.

Performance Metric Description

Silhouette Score Measures cluster separation

(

etter

)

Davies-Bouldin Index Evaluates cluster

compactness (lower is

etter)

Adjusted Rand Index

(

ARI

)

Compares predicted clusters

with actual labels

Normalized Mutual

Informa- tion (NMI)

Measures shared

information between

clusters and labels

These metrics help determine the optimal

clustering approach for identifying high-risk

employees and improving retention strategies (table

3 and figure 8).

Figure 8: Model Performance Metrics.

4.1 Performance Metrics for Employee

Attrition Prediction

Evaluating the performance of machine learning

models in predicting employee attrition is crucial to

ensure reliable results. The following metrics are

used to assess classification performance:

Accuracy

Accuracy =





(13)

Accuracy measures the overall effectiveness of the

model in correctly predicting whether an employee

will stay or leave. A high accuracy score indicates

that the model correctly identifies attrition patterns,

but it may not be sufficient alone in cases of class

imbalance.

Precision

Precision =





(14)

Precision determines how many employees predicted

as leaving (attrition cases) actually left. A high

precision score ensures that the model minimizes

false positives, meaning fewer employees are

incorrectly classified as at-risk.

Recall (Sensitivity)

Recall =





(15)

Recall measures how well the model identifies actual

attrition cases. A higher recall ensures that most

employees who are likely to leave are correctly

detected, reducing the chances of missing critical

attrition risks.

F1-Score

𝐹1 = 2 

PrecisionRecall

PrecisionRecall

(16)

The F1-score is the harmonic mean of precision and

recall, balancing false positives and false negatives.

This metric is especially useful when the dataset is

imbalanced, ensuring a trade-off between predicting

attrition cases correctly while minimizing

misclassifications.

Early Detection of Employee Turnover Risks Using Machine Learning Approaches

125

4.2 Confusion Matrices

Figure 9: Confusion Matrix for Random Forest.

The confusion matrix (figure 9 and 10) illustrates

the number of correct and incorrect classifications for

each model. The color intensity represents the

frequency of predictions, highlighting areas of high

and low accuracy.

Figure 10: Confusion Matrix for SVM.

5 MODEL PERFORMANCE

This study compares the performance of widely used

machine learning models, including Random Forest,

Support Vector Machine (SVM), Logistic

Regression, XGBoost, and Multi-Layer Perceptron

(MLP). The models are evaluated based on essential

classification metrics such as accuracy, precision,

recall, and F1-score to determine their effectiveness

in predicting employee attrition.

As shown in Table 4, Logistic Regression

achieves the highest accuracy (88%), closely

followed by XGBoost (87%) and Random Forest

(85%). Notably, SVM demonstrates the highest recall

(100%), making it particularly effective in

identifying employees likely to leave, though its

precision is comparatively lower. Ensemble-based

models like Random Forest and XGBoost leverage

feature importance, resulting in high precision (88%

and 90%, respectively) and balanced F1-scores (92%

and 93%). The Multi-Layer Perceptron (MLP) also

performs competitively, exhibiting strong recall

(97%) and a reliable F1-score (92%) as shown in

table 4.

Table 4: Performance Comparison of Machine Learning

Models for Employee Attrition Prediction.

Model

Accuracy

(%)

Precision

(%)

Recall

(%)

F1-

Score

(

)

Random

Forest

85 88 96 92

SVM 85 85 100 92

Logistic

Regression

88 88 99 93

XGBoost 87 90 95 93

MLP 85 87 97 92

Figure 11: ROC Curves Comparing Model Performance.

Furthermore, the ROC curves presented in Figure

11 provide a visual comparison of model

performance, illustrating each model’s ability to

distinguish between classes. The curves demonstrate

ICRDICCT‘25 2025 - INTERNATIONAL CONFERENCE ON RESEARCH AND DEVELOPMENT IN INFORMATION,

COMMUNICATION, AND COMPUTING TECHNOLOGIES

126

that Logistic Regression and XGBoost have the most

optimal performance, with areas under the curve

(AUC) close to 1. The analysis highlights the

robustness of these models, making them viable

options for real-world employee attrition prediction

tasks.

6 CONCLUSIONS

Predictive modeling for employee turnover translates

to an understanding of the motivating factors with the

greatest impact on retaining employees and positions

organizations in a proactive manner to engage in

retention practices. Different machine-learning

models are able to demonstrate drivers like

compensation, job satisfaction, work-life balance,

and chances for career opportunities. Predictive

analytics support HR teams in making correct

decisions based on data to bring about engagement

among staff, better working conditions for workers,

and a bigger cut in attrition rates. Early detection of

employees will go a long way in assuring that course

of action can be smoothly initiated to enhance

productivity and stability in the workforce. This

research reiterates the importance of data-driven

workforce management and the potential of

predictive models when it comes to reinforcing

retention efforts. Future developments will entail

incorporating real-time approaches that implement

deep learning to enhance prediction accuracy and

effectiveness in decision-making.

REFERENCES

Arora, A., Acharya, D., & Sinha, A. The classification

performance using logistic regression and support

vector machine (SVM). International Journal of

Engineering and Technical Research, 7(10), 1–4.

https://www.researchgate.net/publication/320409761_

The_classification_performance_using_logistic_regres

sion_and_support_vector_machine_SVM, (2017).

Bhartiya, Namrata, et al.”Employee attrition prediction

using classification models.” 2019 IEEE 5th

International conference for convergence in technology

(I2CT). IEEE, 2019.

Chaurasia, Akansha, et al.”Employee attrition prediction

using artificial neural networks.” 2023 4th International

Conference for Emerging Technology (INCET). IEEE,

2023.

Garg, Umang, et al.”Classification and Prediction of

Employee Attrition Rate using Machine Learning

Classifiers.” 2024 International Conference on

Inventive Computation Technologies (ICICT). IEEE,

2024.

Khekare, Ganesh, et al.”Logistic and linear regression

classifier based increasing accuracy of non-numerical

data for prediction of enhanced employee attrition.”

2023 3rd International Conference on Advance

Computing and Innovative Technologies in

Engineering (ICACITE). IEEE, 2023.

Maharana, Manisha, et al.”Automated early prediction of

employee attrition in industry using machine learning

algorithms.” 2022 10th International Conference on

Reliability, Infocom Technologies and Optimization

(Trends and Future Directions)(ICRITO). IEEE, 2022.

Manikandan, M., et al.”A Novel Methodology Design to

Predict Employee Attrition by Using Hybrid Learning

Strategy.” 2023 International Conference on Innovative

Computing, Intelligent Communication and Smart

Electrical Systems (ICSES). IEEE, 2023.

Mehta, Vimoli, and Shrey Modi. ”Employee attrition

system using tree based ensemble method.” 2021 2nd

International Conference on Communication, Computi

ng and Industry 4.0 (C2I4). IEEE, 2021.

Meraliyev, Bakhtiyor, et al.”Attrition rate measuring in

human resource analytics using machine learning.”

2023 17th International Conference on Electronics

Computer and Computation (ICECCO). IEEE, 2023.

Minaee, S., Kalchbrenner, N., Cambria, E., Nikzad, N.,

Chenaghlu, M., & Gao, J. Deep learning based text

classification: A comprehensive review. arXiv preprint

arXiv:2004.03705, (2020).

Mitravinda, K. M., and Sakshi Shetty. ”Employee attrition:

Prediction, analysis of contributory factors and

recommendations for employee retention.” 2022 IEEE

International conference for women in innovation,

technology entrepreneurship (ICWITE). IEEE, 2022.

Occhipinti, A., Rogers, L., & Angione, C. A pipeline and

comparative study of 12 machine learning models for

text classification. arXiv preprint arXiv:2204.06518.

https://arxiv.org/abs/2204.06518, (2022).

Phadke, Ghanashyam, et al.”Understanding and Managing

Employee Attrition: Strategies for Retaining Talent.”

2023 6th International Conference on Advances in

Science and Technology (ICAST). IEEE, 2023.

Polisetti, Sravanthi, et al.”Stacking Models for Employee

Attrition Prediction: Leveraging Logistic Regression

and Random Forest.” 2024 8th International

Conference on I-SMAC (IoT in Social, Mobile,

Analytics and Cloud)(I-SMAC). IEEE, 2024.

Pratibha, G., and Nagaratna P. Hegde. ”HR analytics: early

prediction of employee attrition using KPCA and

adaptive k-means based logistic regression.” 2022

Second International Conference on Interdisciplinary

Cyber Physical Systems (ICPS). IEEE, 2022.

Rajeswari, G. Raja, et al.”Predicting employee attrition

through machine learning.” 2022 3rd International

Conference on Smart Electronics and Communication

(ICOSEC). IEEE, 2022.

Sharma, Shivansh, and Kapil Sharma. ”Analyzing

Employee’s Attrition and Turnover at Organization

Using Machine Learning Technique.” 2023 3rd

Early Detection of Employee Turnover Risks Using Machine Learning Approaches

127

International Conference on Intelligent Technologies

(CONIT). IEEE, 2023.

Subasri, C. K., & Jeyakumar, V. Comparative analysis of

machine learning algorithms for diabetes prediction

using real-time data set. International Research Journal

of Engineering and Technology (IRJET), 10(3), 1–6,

(2023).

Yadav, Sandeep, Aman Jain, and Deepti Singh. ”Early

prediction of employee attrition using data mining

techniques.” 2018 IEEE 8th international advance

computing conference (IACC). IEEE, 2018.

Yildiz, B., Bilbao, J. I., & Sproul, A. B. A review and

analysis of regression and machine learning models on

commercial building electricity load forecasting.

Renewable and Sustainable Energy Reviews, 73, 1104–

1122. https://doi.org/10.1016/j.rser.2017.02.003,

(2017).

Zheng, D., Hao, X., Khan, M., Wang, L., Li, F., Xiang, N.,

Kang, F., Hamalainen, T., Cong, F., Song, K., & Qiao,

C. Comparison of machine learning and logistic

regression as predictive models for adverse maternal

and neonatal outcomes of preeclampsia: A

retrospective study. Frontiers in Cardiovascular

Medicine, 9,

959649. https://doi.org/10.3389/fcvm.2022.959649,

(2022).

ICRDICCT‘25 2025 - INTERNATIONAL CONFERENCE ON RESEARCH AND DEVELOPMENT IN INFORMATION,

COMMUNICATION, AND COMPUTING TECHNOLOGIES

128