All projects
Shipped

Data Science | School | 1 min read

HumanForYou.

Predicting who is about to quit, from 25 HR variables, with over 80% accuracy. The real work was elsewhere: an ethical audit (GDPR, EU AI Act, disparate impact, equalized odds) to make sure the model judges situations, not people.

  • >80% accuracy
  • Fairness audit

HumanForYou: Turnover Prediction

Technologies: Python, Scikit-learn, Pandas, Matplotlib, Seaborn

Period: Oct - Nov 2025

Overview

Predictive modeling project on HR data (25 variables) to identify employees at risk of leaving.

Methods

  • Data cleaning and exploratory analysis (EDA)
  • Feature engineering: encoding, scaling, feature selection
  • Models compared: Logistic Regression, Random Forest, Gradient Boosting
  • Final accuracy: >80%

Ethical Audit

GDPR / EU AI Act compliance review:

  • Disparate impact analysis by sensitive attribute
  • Equalized odds verification across groups

Stack

Python | Scikit-learn | Pandas | Matplotlib | Seaborn | Jupyter