Salifort Motors Case Study

Employee Retention Analysis and Predictive Modeling

Project Overview

The HR department at Salifort Motors wanted to take initiatives to improve employee satisfaction levels at the company. They collected data from employees but needed help analyzing it to understand what factors contribute to employee turnover.

Business Problem

Employee turnover is time-consuming and expensive for companies. The HR department needed to identify factors that contribute to employees leaving so they could develop targeted retention strategies.

Project Goals

  • Analyze the data collected by the HR department
  • Build a model that predicts whether an employee will leave the company
  • Identify key factors that contribute to employee turnover
  • Provide data-driven recommendations to improve employee retention

Dataset

The dataset contained 15,000 employee records with 10 variables:

  • satisfaction_level: Employee-reported job satisfaction level [0–1]
  • last_evaluation: Score of employee's last performance review [0–1]
  • number_project: Number of projects employee contributes to
  • average_monthly_hours: Average number of hours employee worked per month
  • time_spend_company: How long the employee has been with the company (years)
  • Work_accident: Whether or not the employee experienced an accident while at work
  • left: Whether or not the employee left the company
  • promotion_last_5years: Whether or not the employee was promoted in the last 5 years
  • Department: The employee's department
  • salary: The employee's salary level (low, medium, high)

Methodology

This project followed a standard data science workflow:

Data Preprocessing

  • Data cleaning and handling missing values
  • Feature engineering to create more informative variables
  • Encoding categorical variables for machine learning models
  • Splitting data into training and testing sets

Exploratory Data Analysis

Explored relationships between variables to identify patterns and potential factors contributing to employee turnover. Key areas of investigation included:

  • Relationship between satisfaction levels and turnover
  • Impact of workload (projects and hours) on retention
  • Effect of tenure and promotion on employee decisions to leave
  • Salary level influence on turnover rates

Modeling Approach

Multiple models were built and compared to find the best predictor of employee turnover:

  • Logistic Regression: A baseline model to identify linear relationships
  • Decision Tree: To capture non-linear patterns and feature interactions
  • Random Forest: An ensemble method to improve prediction accuracy

Evaluation Metrics

Models were evaluated using multiple metrics to ensure comprehensive assessment:

  • Accuracy: Overall correctness of predictions
  • Precision: Ability to correctly identify employees who will leave
  • Recall: Ability to find all employees who will leave
  • F1-Score: Harmonic mean of precision and recall
  • ROC-AUC: Model's ability to distinguish between classes

Key Findings & Results

Model Performance

The tree-based models significantly outperformed logistic regression, indicating complex, non-linear relationships between features and employee turnover:

  • Logistic Regression: 83% accuracy, 80% precision, 83% recall, 80% F1-score
  • Decision Tree: 96.2% accuracy, 87.0% precision, 90.4% recall, 88.7% F1-score, 93.8% AUC
  • Random Forest: Slightly outperformed the decision tree model

Key Factors Influencing Employee Turnover

The models identified several critical factors that contribute to employee turnover:

  • Overwork: Employees with too many projects and excessive monthly hours were more likely to leave
  • Satisfaction Level: Low satisfaction was strongly correlated with turnover
  • Time at Company: Employees with around 4 years at the company showed higher turnover rates
  • Evaluation Scores: Both very high and very low evaluation scores were associated with turnover
  • Lack of Promotion: Employees who hadn't been promoted in 5+ years were more likely to leave

Insights

The analysis revealed several important patterns:

  • Employees who worked on too many projects (6+) and put in excessive hours (250+ monthly) were at high risk of leaving
  • Employees with 4 years at the company showed particular dissatisfaction, suggesting a critical retention point
  • High performers who weren't recognized through promotion were likely to seek opportunities elsewhere
  • The relationship between evaluation scores and turnover was non-linear, with both very high and very low scores associated with leaving

Recommendations

Based on the model results and data analysis, the following recommendations were developed to improve employee retention at Salifort Motors:

Workload Management

  • Project Caps: Implement a maximum limit on the number of projects employees can work on simultaneously
  • Hours Monitoring: Track and manage monthly working hours to prevent burnout
  • Workload Distribution: Ensure projects are distributed equitably across teams

Career Development

  • Promotion Pathways: Consider promoting employees who have been with the company for at least four years
  • Career Conversations: Implement regular career development discussions, especially for employees approaching their 4-year mark
  • Skill Development: Provide opportunities for employees to develop new skills and take on different types of projects

Compensation & Recognition

  • Overtime Policy: Either reward employees appropriately for working longer hours or adjust expectations
  • Policy Communication: Ensure employees are familiar with overtime pay policies and work expectations
  • Performance Recognition: Develop a proportionate scale for rewarding employees who contribute more/put in more effort

Company Culture

  • Feedback Mechanisms: Hold company-wide and team-specific discussions to understand and address work culture issues
  • Satisfaction Surveys: Implement regular satisfaction surveys to monitor employee sentiment
  • Work-Life Balance: Promote a culture that values work-life balance and prevents burnout

Next Steps

For future work, consider:

  • Developing models that can predict employee satisfaction and performance scores
  • Conducting cluster analysis to identify distinct employee groups with different needs
  • Implementing a dashboard to monitor key retention metrics in real-time

Jupyter Notebook

The complete analysis, including code, visualizations, and detailed findings, is available in the embedded Jupyter notebook below.

Loading notebook... This may take a moment.