M.S. Applied Data Science - Capstone Chronicles 2025

12

the work and the organization they work for. These questions highlight employee confidence, alignment with job goals, and employee recognition. All strongly influence employee retention. By focusing on these aspects, the organization can develop targeted strategies to strengthen employee

and baseline model, respectively. In addition to the high accuracy, the XGBoost model had a higher weight balance of true positive predictions with an F1-score of 77% Table 3 Model Evaluation Summary

satisfaction, engagement, and reduce turnover. The organization can also use these questions to identify at-risk individuals who are more likely to leave and implement strategies to improve employee satisfaction and engagement. Figure 7 displays the feature importance rankings from the XGBoost Model.

5 Results and Findings Based on the performance of the XGBoost model, there are many questions from the employee survey that can indicate when an employee will decide to leave the organization. In addition to these questions, gender appeared to be a strong indicator of whether an employee would stay or to leave. 5.2 Feature Importance Feature performance uncovers true associations between the target variables and the features (Takefuji, 2025). Based on feature performance, the top survey questions that can help the organization determine if an employee is going to leave or not are: ● Q6: Employees are recognized for providing high-quality products and services. ● Q36: I know what is expected of me on the job. ● Q34: I know how my work relates to the agency’s goals.

Using these questions, the organization can determine strategic ways to connect employees to

106