Contributed by: Arfan Yusuf Shaikh
I am Arfan Yusuf. I graduated from Sydenham College of Commerce in Mumbai and later on joined my Family business of International Human Resource Consultancy. I started with the base as an HR/Operations executive and then developed my skills with core experience in Human Resource Management, currently heading HR and Global Operations.
I work as an HR in the organization, and one of the verticals where I decided to use data science is understanding attrition and increasing retention. The dataset, apart from the attrition tag, contained attributes such as employee’s age, marital status, travel time, total work experience, employee role, total tenure at the organization, total tenure at current role, clocking time, etc. I pre-processed the data retaining relevant variables, and performed exploratory data analysis using a box plot to understand the relationship of attributes with the attrition tag. I identified that employees who travel are more likely to leave than a desk job. Interestingly, unmarried people are more likely to leave than married people are. Over time employees are also more likely to leave.
I developed a logistic regression to predict the probability of attrition on a test dataset and identified it to have an accuracy of 77.3% and AUC of 82.5%. I am going to use my learning from this model to identify the employees who are most likely to leave. This will help me understand the root cause behind their attrition and develop strategies to retain the best talent. Ex- Career development programs and opportunities within the organization for employees who are in the same role/ team for a long time. Strategies identified from the developed solution will definitely help my organization in reducing attrition, thereby reducing the cost of training new employees and increased productivity.
An employee data management and retention solution are helping us to develop the productivity of employee performance. Now, as the performance has increased, our closings have increased, and we are able to close the targets with minimum turnaround time.
Want to learn more of such data science concepts and work on real-world implementation? Join Great Learning’s PGP Data Science and Business Analytics Course and upskill today!
