Proceedings of the SMART–2018, IEEE Conference ID: 44078

2018 International Conference on System Modeling & Advancement in Research Trends, 23rd–24th November, 2018
College of Computing Sciences & Information Technology, Teerthanker Mahaveer University, Moradabad, India

Predicting Employee Attrition using XGBoost

Machine Learning Approach

Rachna Jain1 and Anand Nayyar2

1
Assistant Professor, Department of Computer Science and Engineering (CSE)
Bharati Vidyapeeth’s College of Engineering New Delhi
2
Graduate School, Duy Tan University, Da Nang, Vietrnam
E-mail: [email protected], [email protected]

Abstract—Considering the global competitive scenario, there or any other natural or normal means [1]. There was a time,
is ocean of opportunities for skilled and talented persons in when an employee joins the company, he/she has a thorough
the world, and given a good chance, employees part from one commitment towards workplace and doesn’t leave the job
organization to another. Employee turnover is regarded as for the whole life. But now the face of organizations has
the key issue for all organizations these days, because of its
changed, especially the face of Information technology. The
adverse effects on workplace productivity, and accomplishing
organizational objectives on time. To overcome this problem, situation of organizations and work culture is completely
organizations are now taking support via machine learning different as if off now. Employees of these times are highly
techniques to predict the employee turnover. With high flexible as compared to old aged employees [21,22].
precision in prediction, organizations can take necessary Employee attrition is also defined as the wastage
actions at due course of time for retention or succession rate or total turnover like situation of demand and supply
of employees. Most of the data comes from basic HR based mechanism. Employee attrition creates chronic problems
database systems, which are not highly efficient in prediction in the organization, and it is realized by organizations in
and modeling and these models are not very accurate in data recent times. It also impacts the competitive strength of
models and cannot assist the organizations to take successful
the company due to the ceaseless demand of proficient
decisions. The primary objective of this research paper is
to predict employee attrition i.e. whether the employee is employees and globalization. With the emergence of
planning to leave or continue to work within the organization. next-generation IT companies, new start-up’s, many of
In this paper, we propose a novel model for predicting the employees have shown attrition because of acquiring
Employee Attrition using Machine Learning based approach new learning skills. In the eyes of management, attrition
i.e. XGBoost which is highly robust. In order to validate the is highly expensive and challenging problem to determine
accuracy of the system proposed for Employee Attrition, the concrete solutions. Today’s organizations are refining and
data set is acquired via online database and fetched to the even re-defining various HR policies and practices, but to
system and highly stunning and precision results are shown by some extent, they are successful, but still the rate of attrition
the system with regard to Employee turnover behavior.
is moving up in every quarter or so. [2]
Keywords—Employee Attrition, Turnover, XGBoost, Machine
Learning, Supervised Learning, Supervised Classification A. Attrition-Definition
According to Barron Business Dictionary, the word
I. Introduction “Attrition” is defined as “Normal and Uncontrollable
In recent times, all types of organizations are becoming reduction of a work force because of retirement, death,
very curious and cautious with regard to their market sickness and relocation” [9].
reputation and to gain a competitive edge over others to According to Longman Dictionary of Contemporary
gain huge profits and attain all types of organizational English, Attrition happens “when people leave a company
objectives. Organizations focus on varied HR issues or course of study and are not replaced [10].
and practices. Organizations consider employees as the According to American Heritage Dictionary of English
central resource for everything, so employees must be Language, “Attrition is a gradual, natural reduction
handled with utmost care. It is the primary responsibility in membership or personnel, as through retirement,
of every organization to solve all sorts of employee issues resignation or death” [11].
and provide appropriate solutions and maintain cordial According to Investopedia, “Attrition in business can
relations to boost strong work environment. The serious mean the reduction in staff and employees in a company
concern which affects the organization and even hampers through normal means such as retirement and resignation,
productivity is Attrition. Attrition is regarded as a reduction the loss of customer or clients to old age or to growing out
in the employee’s number via resignation, death, retirement of a company’s target demographic [12].
Copyright © IEEE–2018 978-1-5386-6369-1 113
2018 International Conference on System Modeling & Advancement in Research Trends, 23rd–24th November, 2018
College of Computing Sciences & Information Technology, Teerthanker Mahaveer University, Moradabad, India

B. Reasons for Attrition Section 4 highlights experimental design, analysis and

The following points enlist the reasons for Employee the results of the study conducted. Section 5 concludes the
Attrition: paper with future scope.
•• Unsuitable Behavior: Most of the studies have II. Literature Review
proved that unsuitable behavior creates adverse The turnover of employee can be regarded as deface
effect on productivity by forcing best employees of intellectual capital of the organization. The Literature
to quit the organizations. Rude behavior, back review revolves around the methodologies and techniques
biting, favourism are all the reasons of unsuitable proposed for predicting attrition of employees by several
behavior, forcing the employees to quit current
researchers.
organizations in search for better opportunities
Cotton and Tuttle [4], proposed strongest indicators
[23].
leading to employee attrition or voluntary retirement as age,
•• Imbalance of Work-Life: Organizations have
tenure, pay, job satisfaction and future perceptions. Other
a target to get the work of three or more people
similar researchers have added significant parameters to
performed by a single individual, creating an
attrition like: Gender, Ethnicity, Education, Marital status,
extreme work pressure on the person, which
salary, professional skill enhancement, growth potential
results in de-motivation, longer hours of work as
etc. [3].
well as weekends spoilage. This reason also lay
With the increase in the turnover rate of employees,
strong roots for employee’s attrition [24].
organizations face serious effects on attainting the pre-
•• Employee Misalignment: Sometimes, organizations
defined objectives. It even affects the ongoing work and
hire people, less qualified and unsuitable for the
existing employee’s productivity. Hiring of new employees
job leading to non-synchronization of employees
also wastes the useful financial resources of organization
in work culture and goals achievement which
like training costs, hiring costs etc.
ultimately leads to attrition [25].
Ozolina-Ozola et al. [5] proposed that employee
•• Lack of Decision-Making Ability: Employees
turnover and attribution is explained by employees’
sometimes feel highly insecure with regard to
attitude towards job (by job satisfaction or organizational
their ability to make or contribute in decision
commitment or by both these attitudes). Employee turnover
making. Organizations require employees to
and attrition creates uncertainty among another workforce.
have ownership and empowerment. Empowered
The result of attrition and turnover has the emphasis of
employees have complete independence to make
top management in nearly every organization. It indicates
suggestions and decisions. And if organizations
that the most costly and tough workforce objections facing
don’t boost empowerment which ultimately leads
to job insecurity and results in attrition [26]. organizations is turnover.
•• Inadequate Professional Skills: Sometimes, Hoffman et al. [6] utilized employee’s survey responses
promotion comes to those employees having less regarding their manager to measure people management
capability and even attaining high positions they skills. The study concludes that, the managers with the
cannot become effective leaders. So, ultimately, good people management skills should get rewarded from
those sorts of employees cannot perform the job the company, which would lead to the decrease in employee
well and that impacts job efficiency, which results turnover. Ideally, better people managers should get perks
in forcing the employee to quit the job and this and promotions, which would help with ensuring company’s
also increases attrition in the organization [27]. success. Also, consistent feedback regarding employees
•• Frozen Promotions and Salary Hikes: Over years is required, which keeps the employees engaged and
of research, it is proved that promotion frozen accountable towards their work. The study also analyzed
and non-salary hikes also lay seeds of attrition. that job which requires low skills (for e.g. Truckers, etc.)
Sometimes organizations freeze salary increments is much easier to perform as compared to the jobs which
and promotions and sometimes people find lack require highly skilled workers [29].
of growth professionally and financially, which Ongori et al. [7] stated that organizations should pay
forces the employees to look for opportunities and employees depending on their performances, along with
in turn attrition increases [28]. various incentives such as individual bonus, lump sum
bonus, profits sharing and other benefits. So, if these are put
C. Organization of Paper in place they will decrease employee turnover. The research
Section 2 highlights Literature Review/Related work. also proposed that the knowledge as well as company’s ideas
Section 3 embarks Research Methodology and highlights should be shared widely among the employees too, which
Model proposed with regard to Employee attrition. gives employees the motivation for staying in their present

114 Copyright © IEEE–2018 978-1-5386-6369-1

 Predicting Employee Attrition using XGBoost Machine Learning Approach

organization and become highly responsible for strong Pay and Singh Monetary strategies 2013
performances from the employees. Workforce optimization Compensation such as performance
can also lead to the organization’s success by getting work linked incentives,
done and also establishing accountability. Good working rewards, increment
conditions are also responsible for lower employee attrition in salary helps retain
rates. The study concluded that empowerment of employees employees
would help with the company’s success in retaining its
employees [30]. Ahsana Compensation, 2013
O’Halloran et al. [8] explored how various performance (2013) career opportunity,
related pay (PRP) schemes impact employee turnover. The supervisor support
study tested whether a profit sharing has an impact on effects retention
turnover as compared to other forms of PRP. The study
revealed that there exists a negative relationship between Organizational Mignonac Perceptions of 2013
aggregate measures of PRP and turnover. Table 1 gives Support and disinterested support
Comparison of different Models based on employee Work life decrease employee
balance voluntary turnover.
attrition.
Table 1: Comparison of Different Models Based on
Job Content Bigdeli Internal and external 2013
Employee Attrition
environmental factors,
Factors Authors Research Outcome Year and motivational
factors influence the
employee’s retention.
Flexible work Goud Trust, flexible 2014
arrangements arrangements
employee engagement,
retention leads to
retention & job
engagement
III. Research Methodology
In this research paper, a novel model is proposed
Reward and L.T. Silbert Tangible Rewards 2005 to analyze the employee attrition. All the variables are
Recognition and its outcomes analyzed via some plots and inferences are drawn via
on Organizational exploratory analysis. After exploration some features are
Support. built based on the variables and the final decision is made
to include/exclude some variables.
Promotion and L. Eyster, Strategies to employ & 2008 The Research Design [9] [10] undertaken for the
opportunity for R Johnson retention of previous study was predictive as well as descriptive. The typical
Growth and employees. data science pipeline, called “OSEMN” [11] is followed as
E. Toder demonstrated in figure 1
1. O: Obtaining the data is the first step in solving
Training and Beynon et Training provides 2014 the problem.
development al. employees loyalty and 2. S: Scrubbing or cleaning the data is the next
retention approach. It includes data imputation of invalid or
missing data and fixing column names.
Leadership Chung- Applying Structural 2009 3. E: Exploring the data will be the next right after
Hsiung Equation Model and allow further insight of what our dataset
Fang, to leadership contains. It also looks for any sort of weird or non-
SueTing style, satisfaction, associated data.
Chang, Organization 4. M: Modelling the data provides predictive power
Guan-Li commitment model on whether an employee will leave or not.
Chen 5. I: Interpreting the data to draw necessary
conclusions.

Copyright © IEEE–2018 978-1-5386-6369-1 115

2018 International Conference on System Modeling & Advancement in Research Trends, 23rd–24th November, 2018
College of Computing Sciences & Information Technology, Teerthanker Mahaveer University, Moradabad, India

regularization. Regularization is a method used to

evade overfitting in linear and tree-based models.
3. Enabled Cross Validation: Measuring the
performance of a prediction model on new datasets
based on some set of methods.
4. Missing Values: XGBoost can handle lost values
such that there already exists a trend in the model
for the missing values.
5. Flexibility: It has defined support for objective
functions designed by a user other than regression,
classification etc.
6. Availability: It can be used with languages such as
Python, Julia, R, Java, and Scala.
7. Save and Reload: XGBoost has the feature to
avoid doing the computation again and again, thus
saving the data model and saving time by reloading
it in the future.
8. Tree Pruning [15]: It grows the tree to a maximum
depth and then follow backward pruning till the
Fig. 1: OSEMN Research Design loss function improvement is below a threshold,
The Research conducted considers “Supervised hence different from GBM [16].
Learning” approach of machine learning. The research
IV. Results and Analysis
deals with two main parameters – active (No Attrition) and
terminated (Attrition occurred). A. Experimental Design
The term ‘Boosting’ refers to a family of algorithms, 1). Dataset Description
which converts weak learner to strong learners. It includes
The dataset is acquired from IBM HR Data Analysis.
generating accurate rules of prediction to problems by
(Source:https://www.ibm.com/communities/analytics/
merging irregular and temperately imprecise rules-of-
watson-analytics-blog/hr-employee-attrition/)
thumb. Boosting [12] refers to a meta algorithm the
combines supervised learning to remove bias and converts Table 1: Dataset Attributes with Description
weak learners to strong ones. A classifier is said to be a weak
Name Description
learner if it is less correlated with the factual classification,
Age Numerical Value
whereas a strong learner is a classifier that is subjectively
well-correlated through the factual classification. Attrition Employee Leaving The Company (0=No,
D. XGBOOST 1=Yes)
Business Travel (1=No Travel, 2=Travel Frequently,
XGBoost belongs to boosted tree algorithm and works
3=Tavel Rarely)
on the principle of gradient boosting. As compared to
others, practices a more regularized-model reinforcement Daily Rate Numerical Value - Salary Level
to regulate overfitting and thus improvises performance. It Department (1=Hr, 2=R&D, 3=Sales)
is a fast method consisting of parallel tree construction and
Distance From Numerical Value - The Distance From
planned to be fault tolerant under the distributed setting.
Home Work To Home
The classifier takes data in the form of DMatrix [13]. It is
Education Numerical Value
regarded as an internal data structure used by XGBoost for
both memory efficiency and speed optimization. Education Field (1=Hr, 2=Life Sciences, 3=Marketing,
XGBoost uses gradient boosting (GBM) framework at 4=Medical Sciences, 5=Others, 6=
core. Yet, does better than GBM framework alone. It is used Tehcnical)
for supervised ML problems. Employee Count Numerical Value
The following features were studied and incorporated
Employee Numerical Value - Employee Id
during research: Number
1. Parallel Computing
2. Regularization: This is the major benefit Enviroment Numerical Value - Satisfaction With The
Satisfaction Enviroment
of XGBoost [14]. GBM has no facility for

116 Copyright © IEEE–2018 978-1-5386-6369-1

 Predicting Employee Attrition using XGBoost Machine Learning Approach

Gender (1=Female, 2=Male) 2). Data Preprocessing

In order to perform research, the dataset consists of
lots of attributes which affect employee attrition directly or
Hourly Rate Numerical Value - Hourly Salary
indirectly. As a matter of fact, we had to initially identify
and prune the attributes into a dataset with reduced number
Job Involvement Numerical Value - Job Involvement
of attributes that were relevant for the study.
Job Level Numerical Value - Level Of Job In order to build our model, we trained the model on
a training set and perform validations on the test set. We
Job Role (1=Hc Rep, 2=Hr, 3=Lab Technician, performed all our analysis on the training set and validated
4=Manager, 5= Managing Director, it on our testing set. We divided our dataset into training
6= Reasearch Director, 7= Research (75%) and testing (25%).
Scientist, 8=Sales Executieve, 9= Sales
Representative) 3). Feature Engineering
New features like Tenure per job, Years without change
Job Satisfaction Numerical Value - Satisfaction With The
Job and compa Ratio is calculated on the basis of following
assumptions:
Marital Status (1=Divorced, 2=Married, 3=Single)
1. Tenure per job: People who worked in many
companies for small tenure have a higher chance
Monthly Income Numerical Value - Monthly Salary of leaving that particular company as well to keep
them going.
Monthy Rate Numerical Value - Monthy Rate 2. Years without Change: Changes in job or level
or responsibility keeps the job more exciting and
Numcompanies Numerical Value - No. Of Companies
Worked Worked At less monotonous. Using Promotion, Role and Job
Change as a metric we created a variable to cover
Over 18 (1=Yes, 2=No) the different variants of change.
Overtime (1=No, 2=Yes) 3. Compa Ratio: Compa Ratio [17] is the ratio of
the actual pay of an Employee to the midpoint
Percent Salary Numerical Value - Percentage Increase In of a salary range. The salary range can be that of
Hike Salary his/her department or organization or role. The
benchmark numbers can be an organization’s pay
Performance Numerical Value - Erformance Rating or Industry average.
Rating
To calculate our Compa Ratio at Department Level &
Relations Numerical Value - Relations Satisfaction Organization Level we used the Company pay information.
Satisfaction People with Compa Ratio less than 1, usually feel underpaid
and show more tendency to leave the Organization in search
Standard Hours Numerical Value - Standard Hours of a better pay.
Stock Options Numerical Value - Stock Options
4). Correlation of Attributes
Level The data that we have had a large number of attributes,
Total Working Numerical Value - Total Years Worked but we have used some major attributes in finding out
Years the turnover rate. We have found out many interesting
Training Times Numerical Value - Hours Spent Training relationships among these attributes that led us to our goal of
Last Year finding the turnover rate and in which year the turnover rate
Work Life Numerical Value - Time Spent Bewtween touched its peak. In our data, we have shown a correlation
Balance Work And Outside between attributes such as how many years an employee
spent in a company, how many years an employee spent
Years At Numerical Value - Total Number Of Years
in a company with current manager and how many years
Company At The Compnay
spent in the company since the last promotion. We have
Years In Current Numerical Value -Years In Current Role also shown the correlation between the level of job or
Role service an employee is doing and monthly income of the
Years Since Last Numerical Value - Last Promotion employee. We have also considered the relation between
Promotion the attributes like percent hike and the performance rating
of an employee. We have also found out the correlation
Years With Numerical Value - Years Spent With
between attributes such as number of years spent by an
Current Manager Current Manager
employee under the current manager, the level of the job

Copyright © IEEE–2018 978-1-5386-6369-1 117

2018 International Conference on System Modeling & Advancement in Research Trends, 23rd–24th November, 2018
College of Computing Sciences & Information Technology, Teerthanker Mahaveer University, Moradabad, India

and percentage of hike in salary. So, we have used a number 4. D

​ epartment: There were less number of people
of attributes and correlations among them to find out the attired from the HR department. The reason
turnover rate of a company in a certain period of time. behind this is that there were the low proportion of
5). Binning HR employees in the organization
We needed to categorize the variables which can collect 5. Job Involvement: The ratings were given as
the impact making groups and make more sense as many of ‘low’, ‘medium’, ‘high’ and ‘very high’. The study
our variables are either continuous in nature or have lots of showed that the majority of employees who left
discrete values which peak at certain points. While doing the organization were either highly involved or
these transformations, we ensure that we do the same for low involved in their jobs.
our Testing set also. 6. Job Satisfaction: Looking at the job satisfaction,
B. Analysis the higher levels of attrition were observed in
The research study concluded that there are certain lower job satisfaction levels.
variables that influence the turnover rate of an organization.
The variables highly influencing the attrition are enlisted
as follows:
1. Age: Majority of employees that were around 30
years old had left their organization.

Fig. 4: Attrition W.R.T Job Satisfaction

7. Marital Status: Attrition rates were higher if

an employee was unmarried and lowest for the
Divorced employees.

Fig. 2: Attrition W.R.T Age

2. Gender: In our study, we found out that a large
number of attired individuals were male and it is
because 61 percent of employees in the dataset
were male.

Fig. 5: Attrition W.R.T. Maritial Status

8. Monthly Income: Our study showed that if the

monthly income of an employee is low, then their
chances of leaving the company are high. It might
be due to dissatisfaction with the income compared
Fig. 3: Attrition Percentage in Data to the effort they are putting out.
3. Distance from home: A large number of the 9. Y
​ ears at company: The study showed that large
employees left the organization just because the number of newcomers, left the organization
office was close to their home. This is in contrast which sidelined the recruitment efforts of the
to the normal assumptions. organization.
118 Copyright © IEEE–2018 978-1-5386-6369-1
 Predicting Employee Attrition using XGBoost Machine Learning Approach

10. Y​ ears in current role: Majority of employees

who worked for less than a year had left their
organization. This might occur because they were
offered a different role in a different company.
11. ​Years since last promotion: Majority of
employees who were promoted recently, quit their
jobs.
12. ​Years with current Manager: Attrition rate is
higher when the current manager of an employee
is replaced with the new ones.

Fig. 6: Confusion Matrix and other Statistics

V. Conclusion
In this paper, a machine learning approach for predicting
employee attrition is presented in this paper. The most
significant drawback of existing organization’s data models
and database is that, they contain lots of redundant data and
predicting something with precision is quite challenging.
We implemented a precision model for predicting employee
attrition using XGBoost based machine learning technique.
XGBoost is regarded as a superior algorithm in terms of
efficient memory utilization, high accuracy and low running
times. It is simply highly robust and scalable technique to
handle all sorts of noise from huge data sets and convert
the data into a ready acceptable form for precision results.
The model presented in this paper has very low rate less
than 30% and the accuracy touches almost to 90%. Because
of these reasons, XGBoost technique is recommended on
top priority manner for employee turnover prediction to
successfully enable the organization to take preventive
action in due course of time.
A. Future Scope
In the near future, we like to implement the proposed
model in real-world organizations to enable organizations
to learn the employee turnover reasons. The research
After the implementation of the model, we concluded would go in the direction to make this model a “Predictive
that a total of 14 factors influence the attrition rate more Mode” and solve various issues, i.e. Advanced ones not
than any other factors. After using the baseline decision tree predicting- “Who is going to Leave?” but also “Why the
models that had maximum accuracy up to 83 percent, we Employees are doing turnover?”. The model will become
decided to use a more advanced approach. For the same, we more accurate, scalable and ready to implement as such in
considered glmboost [18] and XGBoost [19] [20] techniques top IT organizations HR departments.
for our model. Our model, based on XGBoost worked the References
best for us with a decent specificity rate (> 50%) and a very [1] Nappinnai, M.V., & Premavathy, N. (2013). Employee attrition and
low error rate (< 30 %). It had an improved accuracy over retention in a global competitive scenario. International Journal of
Research in Business Management (IMPACT: IJRBM), 1(6), 11–14.
the baseline model which increased up to 89% as shown [2] Adhikari, A. (2009). Factors affecting employee attrition: a multiple
in Fig.6. The model was quite robust over its counterpart regression approach. IUP Journal of Management Research, 8(5),
glmboost in terms of accuracy as well as error rate. 38.

Copyright © IEEE–2018 978-1-5386-6369-1 119

2018 International Conference on System Modeling & Advancement in Research Trends, 23rd–24th November, 2018
College of Computing Sciences & Information Technology, Teerthanker Mahaveer University, Moradabad, India
[3] M. Young, The Technical Writer’s Handbook. Mill Valley, CA: [20] Abbasi, S.M., & Hollman, K. W. (2000). Turnover: The real bottom
University Science, 1989. line. public Personnel management, 29 (3), 333–342.
[4] J.L. Cotton and J. M. Tuttle, “Employee turnover: A meta-analysis [21] Farooq, S., & Hanif, N. (2013). A Descriptive Study of Intrinsic and
and review with implications for research”, Academy of management Extrinsic Motivational Factors and Their Role in Employee Retention
Review, 11(1), 55–70, 1986. in Banking Sector (Lahore) Pakistan. Int. Journal of Innovative and
[5] Ozolina-Ozola, I. Reducing Employee Turnover in Small Business. Applied Finance–IJIAF, 1(1).
[6] Hoffman, M., & Tadelis, S. (2018). People Management Skills, [22] Beynon, M.J., Jones, P., Pickernell, D., & Packham, G. (2014).
Employee Attrition, and Manager Rewards: An Empirical Analysis Investigating the impact of training influence on employee retention
(No. w24360). National Bureau of Economic Research. in small and medium enterprises: a regression‐type classification and
[7] Ongori, H. (2007). A review of the literature on employee turnover. ranking believe simplex analysis on sparse data. Expert Systems.
[8] O’Halloran, P.L. (2012). Performance pay and employee turnover [23] Young, J.A., Stone, J., Aliaga, O., & Shuck, B. (2013). Job
Journal of Economic Studies, 39(6), 653–674. Embeddedness Theory: Can It Help Explain Employee Retention
[9] [9] Latha, K.L. (2013). A study on employee attrition and retention Among Extension Agents?. Journal of Extension, 51(4), 4FEA7
in manufacturing industries. BVIMSR’s Journal of Management [24] Goud, P.V. (2014). Employee Retention for Sustainable
Research (BJMR), 5(1), 1–23. Development . International Journal of Innovative Technology &
[10] Ajit, P. (2016). Prediction of employee turnover in organizations Adaptive Management (IJITAM), 1(5), 10–16.
using machine learning algorithms. algorithms, 4(5), C5. [25] Zhang, Y. (2013). Research of Employee Retention on Job
[11] Jahan, S. S. (2014). Human resources information system (HRIS): Embeddedness Perspective . International Conference on Advances
a theoretical perspective. Journal of Human Resource and in Social Science, Humanities, and Management, Atlantis Press,
Sustainability Studies, 2 (02), 33. 343–346.
[12] Alao, D., & Adeyemo, A.B. (2013). Analyzing employee attrition [26] Pranita S. Jumade & Dr. ShubhadaKulkarni (2015),―A Study of
using decision tree algorithms. Computing, Information Systems, Talent Management Strategies in I.T. Sector in India‖, International
Development Informatics and Allied Research Journal, 4. Journal Of Multifaceted And Multilingual Studies, Vol2 No1, pp.
[13] Saradhi, V.V., & Palshikar, G. K. (2011). Employee churn prediction. 1–8.
Expert Systems with Applications, 38(3), 1999–2006. [27] Sultana Nazia&Bushra Begum (2013), “EMPLOYEE RETENTION
[14] Glu, Z. Ö. K. (2014). Employee Turnover Prediction Using Machine PRACTICES IN INDIAN CORPORATE – A STUDY OF SELECT
Learning Based Methods (Doctoral Dissertation, Middle East MNCs‖, International Journal of Engineering and Management
Technical University). Sciences, Vol 4 N0 3, pp. 361–368.
[15] Srivastava, D.K., & Nair, P. (2017, March). Employee Attrition [28] Mohan.S., & Dr. Muthuswamy.P.R.,(2015) “A study on employee
Analysis Using Predictive Techniques. In International Conference retention in BPO Sector with special reference to Coimbotore”
on Information and Communication Technology for Intelligent International journal of Informative and Futuristic Research, Vol 2,
Systems (pp. 293–300). Springer, Cham. Issue 6.
[16] Holweg, M., Davies, P., & Podpolny, D. (2009). The competitive [29] Bidyut B.N., & Dr. Mukulesh B.,(2015) “Factors affecting
status of the UK automotive industry. Buckingham: Picsie Books. Employee’s Retention in Automobile Service Workshops of
[17] DeNisi, A., & Griffin, R. (2015). HR3. Cengage Learning. Assam:An Empirical study” The SIJ Transactions on Industrial,
[18] Memoria, C.B., & Gankar, S.V. Personnel Management Text and Financial & Business Management (IFBM), Vol. 3, No. 1.
Cases, (2002). [30] Karatepe, O.M. (2013) High—Performance Work Practices, Work
[19] Dowling, P.J., & Welch, D. E. (1988). International human resource Social Support and Their Effects on Job Embeddedness and Turnover
management: An Australian perspective. Asia Pacific Journal of Intentions. International Journal of Contemporary Hospitality
Management, 6 (1), 39-65. Management, 25, 903–9

120 Copyright © IEEE–2018 978-1-5386-6369-1