HR TalentScope
ML Analytics Platform for Job-Change Prediction
Built an end-to-end HR analytics platform that predicts employee job-switching behavior using machine learning through a unified Flask dashboard.
System Architecture
End-to-end workforce analytics and prediction flow documented in HR TalentScope.
Workforce records
19,158 labelled training records enter a repeatable analytics and prediction workflow.
EDA + data quality
Ten analytical modules expose workforce distributions, missingness, and job-change patterns.
Feature engineering
Categorical and numeric variables are encoded, scaled, and prepared without leaking target information.
Five-model comparison
Classification candidates are compared through consistent validation and probability-based scoring.
Flask decision workspace
Users inspect trends, submit records, review probabilities, and export prediction outputs.
Job-change probability analysis paired with interactive HR analytics.
Project Overview
HR TalentScope is an end-to-end machine learning analytics platform designed to identify employees who are actively seeking new job opportunities before they leave an organization.
The system combines exploratory data analysis, feature engineering, predictive modeling, and interactive visualization into a single workflow for HR decision support.
The platform processes a dataset containing over 19,000 candidate records, analyzes workforce characteristics, and predicts job-change probability using multiple supervised learning algorithms.
The application follows a complete five-stage pipeline: data loading and profiling, exploratory data analysis, feature engineering and preprocessing, model training and evaluation, and probability-based prediction and export.
The final platform provides an interactive dashboard where users can analyze workforce data, compare model performance, interpret results, and generate job-change probability predictions through a streamlined interface.
Problem
HR teams need to identify employees or candidates with higher job-change probability before churn risk becomes visible through traditional reporting.
Solution
Built a Flask-based ML analytics dashboard combining automated EDA, preprocessing, model comparison, and probability-based prediction exports.