Portfolio.

user image

Advanced Data Analysis for HR Attrition Using Python

Cover Image for Advanced Data Analysis for HR Attrition Using Python

Predicting Employee Attrition: A Data-Driven Approach with IBM HR Analytics (Follow link below to the dashboard)

Introduction

Employee attrition is a costly challenge, affecting productivity, morale, and organizational stability. Using IBM’s HR Analytics dataset, this project uncovers hidden patterns in employee turnover and delivers actionable insights to help businesses retain top talent and optimize workforce strategies.

Key Objectives

  1. Diagnose Attrition Drivers: Identify demographic, role-specific, and behavioral factors linked to turnover.

  2. Quantify Correlations: Analyze relationships between salary, job satisfaction, workload, and attrition risk.

  3. Predict Turnover: Build ML models to flag at-risk employees with high accuracy.

  4. Recommend Interventions: Propose targeted HR strategies to reduce attrition.

Methodology

1. Data Exploration & Cleaning

  • Processed 1,470 employee records with features like:

  • Demographics (age, gender, education)

  • Job-related (role, salary, overtime, promotions)

  • Behavioral (job satisfaction, work-life balance ratings).

  • Addressed missing values, outliers, and categorical encoding.

2. Exploratory Data Analysis (EDA)

- Key Findings:

  • - Employees working overtime were 2.3× more likely to leave.

  • - Low satisfaction scores (<3/10) correlated with 68% higher attrition.

  • - Sales and R&D roles had the highest turnover rates (25% of departures).

3. Predictive Modeling

  • Trained Random Forest, LightGBM, and SVM models to predict attrition risk.

  • Best Model: SVM (Recall: 99%) — accurately flagged 99% of at-risk employees.

  • Handled Imbalance: Applied SMOTE to improve minority-class (attrition) prediction.

Strategic Recommendations

1. Target High-Risk Groups:

- Offer retention bonuses or flexible schedules to overtime employees.

- Address burnout in Sales/R&D teams with workload audits.

2. Boost Satisfaction:

- Launch mentorship programs for employees with <2 years of tenure.

- Tie promotions to career development plans.

3. Proactive Alerts:

- Integrate the model into HR systems to flag at-risk employees in real time.

Conclusion

By combining exploratory analytics and machine learning, this project empowers HR teams to predict attrition before it happens and implement evidence-based retention strategies. The result? A more engaged, stable workforce and lower turnover costs.

Check it out!