Client Name

Habib Bank Limited

Faculty Advisor

Dr. Usman Nazir

SBS Thought Leadership Areas

Behavioural Studies

SBS Thought Leadership Area Justification

Academically, this project falls within the Behavioral Studies thought leadership area of IBA’s School of Business Studies. It leverages behavioral data, such as performance and value alignment metrics, in combination with advanced machine learning techniques, to draw insights into employee decision-making and workplace dynamics. This project directly engages with fundamental questions central to behavioural inquiry: What motivates individuals to stay or leave? How do workplace conditions, organisational culture, and personal values shape decisions in complex systems like banks? By analysing attrition not only as a technical problem but also as a behavioural signal, the project exemplifies the kind of interdisciplinary research that Behavioural Studies promotes. The use of real-world data to explore the links between perception, performance, and organisational action reflects the core mission of the thought leadership area, to understand human behaviour in business contexts and use those insights to inform evidence-based interventions.

Aligned SDGs

GOAL 8: Decent Work and Economic Growth

Aligned SDGs Justification

This project contributes directly to the Sustainable Development Goal 8 (Decent Work and Economic Growth), which emphasizes the need for inclusive, productive, and sustainable employment. By enabling a more nuanced understanding of attrition and supporting data informed retention strategies, the system designed through this ELP empowers HBL to foster a healthier and more responsive workplace. Predictive attrition analysis allows organizations to intervene earlier, reduce involuntary turnover, and align HR policies with employee needs, core objectives of SDG 8. Moreover, the ability to identify organisational causes behind attrition, such as policy dissatisfaction or poor supervisory relationships, supports the creation of fairer, more inclusive working environments. This aligns directly with SDG 8's targets of ensuring labour rights and promoting safe and secure workplaces for all employees. In doing so, the project strengthens institutional resilience by addressing human resource vulnerabilities before they escalate into broader performance issues. Ultimately, reducing preventable attrition not only improves employee well-being but also contributes to long-term operational stability, which is essential for sustainable economic growth.

NDA

Yes

Abstract

This Experiential Learning Project (ELP), undertaken by a team of final-year undergraduate students at the Institute of Business Administration (IBA), Karachi, was conducted in collaboration with Habib Bank Limited (HBL). The primary objective of the project was to assist HBL in developing a machine learning-based system to support employee attrition analysis. The deliverable comprised a fully operational Python-based code capable of aiding in predictive analysis for attrition using historical data, thereby enabling HBL to identify potential retention risks and strengthen workforce management strategies. The data for the project was provided by HBL under strict confidentiality agreements and was accessed exclusively on-site due to the sensitive nature of employee information. The dataset included records of employees who had exited the organization during 2023, 2024, and early 2025. Key variables included demographic information, job function, grade, performance and value ratings, and the reason for exit. Importantly, the dataset did not include records of retained employees, necessitating the use of an unsupervised machine learning approach. The team employed the Isolation Forest algorithm, an anomaly detection technique designed to identify data points that deviate significantly from the norm. The model was trained on the cleaned and standardized dataset and configured to flag the top 10% most anomalous exits based on their profiles. These anomalies were interpreted as employee exits that were unusual in comparison to others in the dataset. Integration of the model with preprocessing logic allowed the code to operate independently on future internal datasets without requiring manual data cleaning. A post hoc statistical analysis was conducted on the anomaly scores using one-way ANOVA and Tukey’s HSD test to determine whether the ‘Leave Reason’ field was significantly associated with anomaly status. The results revealed that departures tied to organizational issues, such as internal grievances or dissatisfaction with policies, were more likely to be classified as anomalous. These findings suggest that certain categories of attrition may serve as early warning signals of internal dysfunction or policy gaps. All statistical assumptions required for valid ANOVA were tested and satisfied, adding robustness to the results. Furthermore, the project addressed the ethical considerations of using machine learning in HR contexts, including transparency, bias, interpretability, and responsible data handling. Overall, the project demonstrates that data-driven approaches can enhance attrition diagnostics, even when constrained by partial datasets. The final codebase allows HBL to apply anomaly detection to future exit data, enabling proactive responses to atypical departures. Beyond technical success, the project also contributes to Sustainable Development Goal 8 (Decent Work and Economic Growth) by supporting more strategic, fair, and evidence-based HR practices. It aligns with the Behavioral Studies thought leadership area at IBA by combining psychological theories, statistical modeling, and artificial intelligence to address real-world organizational challenges.

Document Type

Restricted Access

Document Name for Citation

Experiential Learning Project

Download

Share

COinS