Predicting Employee Attrition with a Comprehensive Machine Learning Approach: Utilizing Ensemble Methods and Hyperparameter Optimization

Sridhar, Sreejith (2024) Predicting Employee Attrition with a Comprehensive Machine Learning Approach: Utilizing Ensemble Methods and Hyperparameter Optimization. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (1MB) | Preview
Preview	PDF (Configuration Manual) Download (994kB) | Preview

Abstract

Employee attrition poses significant challenges for organizations, leading to high recruitment costs, the loss of key personnel, and diminished employee morale. Traditional attrition prediction approaches like surveys and interviews lack the data and timeliness needed for effective intervention. This work develops an effective employee turnover prediction model using machine learning to overcome these constraints. The study seeks to discover attrition factors and create a reliable forecasting model. The CRISP-DM structure guides the research from business needs to data preparation. The dataset was retrieved from Kaggle and thoroughly pre-processed to remove duplicates, encode category variables, and normalise numerical characteristics. The Research tested Logistic Regression, Decision Tree, K-Nearest Neighbours (KNN), Support Vector Machine (SVM), Random Forest, and AdaBoost. Ensemble approaches, especially the Voting Classifier, improved prediction accuracy. GridSearchCV hyperparameter adjustment improved model performance. The optimised Voting Classifier surpassed others in accuracy, precision, recall, F1-score, and ROC AUC, scoring Accuracy of 85.03% and ROC score of 0.845. Overtime, experience, stock option level, and distance from home greatly affect attrition, according to feature importance analysis. Using Streamlit, this research provides an interactive tool for analysing attrition risk and early intervention alternatives. To improve attrition prediction, advanced feature engineering, time-series analysis, and ensemble model interpretability will be investigated in future work.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Horn, Christian UNSPECIFIED
Uncontrolled Keywords:	Employee Attrition; Machine Learning; Prediction Model; CRISP-DM; Feature Importance; Voting Classifier; GridSearchCV; Hyperparameter; Ensemble
Subjects:	Q Science > QA Mathematics > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning H Social Sciences > HD Industries. Land use. Labor > Issues of Labour and Work > Staff Turnover
Divisions:	School of Computing > Master of Science in Data Analytics
Depositing User:	Ciara O'Brien
Date Deposited:	26 Aug 2025 11:14
Last Modified:	26 Aug 2025 11:14
URI:	https://norma.ncirl.ie/id/eprint/8636

Actions (login required)

View Item