Case Study 2 - Doing Data Science Fall 2023
In the introduction phase, the project embarked on a journey to address the pivotal question posed by DDSAnalytics: how to leverage data science to predict employee turnover (attrition). The overarching goal was to provide actionable insights for talent management and retention. This section introduces DDSAnalytics as a leading player in talent management solutions and sets the stage for the subsequent detailed analyses and model developments.
A critical step in the project involved meticulous data cleaning and exploratory data analysis (EDA). The dataset underwent thorough scrutiny, ensuring the integrity and quality of the information. Exploratory analyses delved into key features such as MonthlyIncome, JobLevel, and WorkLifeBalance to uncover trends and patterns that would serve as the foundation for subsequent modeling efforts.
To enhance the interpretability of complex relationships within the data, interactive RShiny Apps were developed. These apps provided a user-friendly interface for dynamic exploration of key variables, offering executives at DDSAnalytics a hands-on tool for visualizing and understanding the nuances of attrition factors.
The core of the project involved the development of a robust predictive model for attrition. Leveraging a combination of statistical tests and the Naive Bayes algorithm, the model achieved a delicate balance between sensitivity and specificity. This section outlines the methodologies employed, the rationale behind the chosen approach, and the model's performance in predicting employee turnover.
In parallel with attrition prediction, a linear regression model was crafted to predict MonthlyIncome. The model showcased its effectiveness with a Root Mean Square Error (RMSE) of 1400, underlining its utility in providing insights into the salary dynamics within the organization.
Statistical analyses were conducted to discern the key factors contributing to attrition. The identification of the top three factors was informed by rigorous experimentation, ensuring that the chosen variables held significance in predicting employee turnover. This section sheds light on the crucial findings that drive talent management strategies.