Salifort Motors Employee Retention Prediction


End-to-End ML Project

This project focuses on predicting employee attrition at Salifort Motors, a fictional multinational alternative energy vehicle manufacturer. It covers ETL, EDA, Feature Engineering, Model Training, Hyperparameter Tuning, and performance evaluation.


Data Source:

The dataset contains 14,999 anonymized employee records with the following features:

  • Satisfaction level: Employee satisfaction score
  • Last evaluation score: Performance evaluation score
  • Number of projects: Total projects handled
  • Average monthly hours: Average hours worked per month
  • Time spent at the company: Tenure in years
  • Work accident history: Whether the employee had a work accident
  • Recent promotion status: Whether the employee was recently promoted
  • Department: Department of the employee
  • Salary level: Low, medium, or high
  • Attrition (Target Variable): Whether the employee left the company

Phase 1: Data Cleaning

The dataset was cleaned by handling missing values, encoding categorical variables, and normalizing numerical features. The cleaned dataset was saved for further analysis.

Phase 2: Exploratory Data Analysis (EDA)

Key insights from EDA include:

  • High and low extremes in average monthly hours are associated with higher attrition.
  • Longer tenure and lack of promotions correlate with higher turnover rates.
  • Low salary groups and certain departments show elevated attrition.
  • Satisfaction level is a strong inverse predictor of attrition.

Phase 3: Modeling and Evaluation

Several models were developed and evaluated. The final model, a tuned Random Forest Classifier, achieved the following metrics:

  • Accuracy: 97%
  • Recall (Attrition Class): 96%
  • F1-Score (Attrition Class): 93%
The model was saved using joblib for future use.

Link to GitHub Repository: Salifort Motors Employee Retention Prediction