HR Analytics:
Employee Attrition

💼 About the Project

Human resources (HR) are a crucial aspect of any organization, as they deal with the management of an organization's most valuable resource - its employees. With the increasing amount of data being generated in the workplace, HR analytics has become an essential tool for organizations to better understand their workforce and make data-driven decisions.

This project aims to develop a human resource data analysis system that provides insights into various aspects of an organization's HR operations. The system will use machine learning algorithms and data visualization tools to analyze HR data, including employee demographics, performance metrics, compensation, and benefits, among others.

Tools used

  • python 3.8
  • Tableau

Project Repository

GitHub Repository

🚩 Problem Statment

With the increasing amount of data being generated in the workplace, it is challenging HR departments to connect relationship between the data and the cause of HR event such as 'attrition' without deeper analysis of their HR data.

🧐 Research Questions

The project aims to answer the listed questions

  • What features people leaving the company have in common?
  • What is the most important infomation from the dataset that affects attrition rate?
  • What improvement points can the organization make to better manage their employees?

    • 📈 Workflow

    1. Exploratory Data Analysis
    2. Findings
    3. Data Modeling
    4. Suggestions

    🗺️ Exloratory Data Analysis

    Visualization of the data will unlock the hidden insights from the dataset
  • Synthetic HR data from IBM
  • 16.1 % of employee attrition rate is observed
  • Younger employees tend to have higher attrition rate
  • Men tend to have higher attrition rate
  • Marital status does not shw significant relationship
  • Lower monthly income shows higher attrition rate
  • Employee overtime status effects on the attrition
  • Years with company also shows a clear trend
  • Frequent business travel shows high attrition

    • 🧐 Findings

    From EDA, what insight can we find?
  • Generalists vs specialists in education
  • Generalists vs specialists in job role
  • Sales Representitive shows the lowest avg income
  • Sales Representitive shows the lowest avg job involvement
  • Survey data including performance rating does not show any information

    • 🔮 Prediction Modeling

  • BOX-COX transformation to deal with outliers
  • Correlation chart identifies the feature relationship
  • Feature importance from the correlation chart, top (+), bottom (-)
  • Overtime shows the most importance
  • While performance rating the least
  • With logistic Regression model, we are able to predict the attrition with 90% Accuracy

    • 🚀 Key Insights & Limitation

  • Overtime plays a critical role in employee attrition.
  • Investing resource in generalist (HR, Sales, Marketing, etc) positions will benefit the organization.
  • Having a large talent pull of potential sales representative candidates would be ideal to manage sales talent.
  • Performance reporting system is not properly functioning, consider re-designing the overal survey system.