Attention to Retention: Exploring the Dynamics of Data Scientist Job Mobility

High turnover rates are a significant challenge in the data science industry, leading to concerns about talent retention and long-term growth. This study tackles the pressing challenge of predicting job change behavior among data scientists using machine learning to inform effective retention strategies. We analyze the “HR Analytics: Job Change of Data Scientists” dataset on Kaggle, employing a rigorous pipeline that addresses categorical data, missing values, and class imbalance through techniques such as Label Encoding, KNN Imputation, and up-sampling. Seven machine learning models are evaluated, with Extreme Gradient Boosting emerging as the frontrunner, attaining a test score of 84.4% and a noteworthy F1 score of 84.5%. To delve deeper into model interpretability, we leverage feature importance analysis, revealing key insights into factors influencing job change. By proactively identifying individuals at risk and implementing targeted interventions based on predictive insights, organizations can foster a more engaging and fulfilling work environment, leading to enhanced retention rates, employee satisfaction, and overall lower employee attrition.