HR Analytics to Track
Employee Performance
In today's competitive business environment, HR analytics is a powerful
tool for monitoring and optimizing employee performance. This research
explores its role in performance management, discussing benefits,
challenges, and ethical considerations. We investigate key performance
indicators (KPIs) relevant to employee performance tracking and their
alignment with organizational goals. AI and ML strengthen analytical
power. The study offers insights for organizations interested in using data-
based approaches in performance management.
by Sakshi Sonavane
Literature Survey on HR Analytics
Machine Learning for Predictive Analytics in Remote Work
Turnover Analysis Manufacturing Performance Evaluation
Karimi & Vilijavan (2024) discuss using Zhao, Lin, & Wang (2023) explore deep Singh, Mehta, & Patel (2023) analyze
ML algorithms to predict employee learning for predictive analytics. They remote employee performance. They
turnover. This helps in improving HR aim to identify key turnover factors in use time-series analytics to identify
decision-making by identifying manufacturing industries to reduce trends and productivity drivers in
potential departures. attrition. remote settings.
HR Analytics and Attrition
Understanding HR Predictive Analytics Impact on Growth
Attrition
HR managers use data analytics to Employee attrition affects the overall
HR attrition refers to employees predict employee departures. This growth strategies of companies.
leaving the company, which can be a helps organizations take proactive Effective HR analytics can mitigate
critical issue. High attrition rates lead measures to retain valuable these negative impacts.
to increased costs in recruitment and employees.
training.
Working of HR Analysis
Data Collection
Gather employee data from various sources.
Data Analysis
Analyze relationships between different features.
Attrition Prediction
Predict employee attrition status.
Policy Implementation
Implement new policies to reduce attrition.
Methodology and Data
Sources
Data Sources Data Pre- Machine
processing Learning
The IBM HR employee
Techniques
dataset was used for A three-stage system
model building and (preprocessing, Techniques like Logistic
evaluation. This dataset processing, and post- Regression, Random
includes a variety of processing) was Forest, SVM, and
employee-related employed to predict Decision Tree were
features. employee attrition used for employee
effectively. attrition prediction.
Google Collab, Jupyter Notebook, Power BI, Dash, Flask and Plotly were used
to implement the data visualizations.
Modelling and Analysis: Logistic Regression
Logistic Regression Overview Application in Attrition Analysis
Logistic regression is used for binary classification tasks. It It predicts whether an employee will stay or leave, using
predicts the probability of a categorical outcome based on predictors like age, tenure, job satisfaction, and performance
predictor variables. metrics.
The model exhibits high accuracy in identifying employees likely to remain. However, it shows limitations in predicting those
likely to leave. It effectively identifies employees staying with high precision and recall, but its performance in detecting
employees leaving is weaker.
Modelling and Analysis:
Machine Learning
1 Random Forest 2 Support Vector
Machine (SVM)
An ensemble learning
method combining Aims to find a hyperplane
multiple decision trees to that distinctly separates
form a stronger learner, data points into different
reducing overfitting. classes, maximizing the
margin between them.
3 Decision Trees (DT)
Approximates a target function as a tree of "if-then" rules,
improving human readability and handling both categorical and
numerical data.
Results and Discussion
Metric Value
Male Employees 60%
Female Employees 40%
R&D Department Employees 65.4%
Sales Department Employees 30.3%
HR Department Employees 4.29%
IBM has 60% male and 40% female employees. Most employees work in
R&D (65.4%), followed by Sales (30.3%) and HR (4.29%). Attrition
decreases as age increases. Younger employees tend to leave more often.
Conclusion and Key
Findings
Employee attrition was analyzed using machine learning models: Logistic
Regression, Random Forest, and SVM. The organization experiences a
16.12% attrition rate, with gender disparities and age-related trends.
Key findings show that frequent travelers have higher attrition, while the
R&D department has the lowest attrition rate. Monthly Income, Age, and
Over Time were identified as top contributors to attrition.
Addressing overtime concerns could potentially reduce attrition rates
within the company. We express our sincere gratitude to our project
guide, Prof. Swati Bhoir, for their guidance and support.
