← Back![]()
Employee attrition risk factors
Insight dashboard + decision tree to prioritise retention
Problem
Understand which factors (education, distance to office, salary, satisfaction, gender, department) are associated with employee resignation and propose retention actions.
Data
100 employees (18 resignations) in «Employee Data (1).xlsx».
Methods
- Attrition rate by category and statistical ranking (chi²)
- Decision tree with one-hot encoding, confusion matrix and cross-validation
- Simplified 2-variable tree compared to a baseline (DummyClassifier)
- Dashboard with salary × satisfaction and department × gender cross-tabs
Results
- Overall attrition: 18% (n = 100).
- Low salary + low satisfaction: 75% attrition (n = 4, warning signal); high salary + high satisfaction: 0% (n = 18).
- Low satisfaction: 28.57% attrition vs 15.63% (medium) and 14.89% (high).
- Distance does not explain attrition: Near 19.4% (n = 67) vs Far 15.2% (n = 33).
- Men in Sales: 27.8% (n = 18), the highest department × gender cell.
- The simplified tree identifies satisfaction and salary level as the main criteria; in the initial test it caught every resignation, with many false alarms. With 100 records it is framed as exploratory, not predictive.
Gallery
Tools
Pythonpandasscikit-learnDatabricksLooker Studio