Data Normalization Using Quantile Isolation Mapping And Feature Selection For Employee Attrition Prediction

Employee attrition prediction is a critical task for organizations aiming to retain talent and maintain workforce stability. This study focuses on two essential phases of the prediction process: data normalization using Quantile Isolation Mapping (QIM) and feature selection employing Partitioned Binary Bat with Optimized Random Forest (PBB-ORF). The initial phase involves data normalization through QIM, a robust technique designed to handle outliers effectively. By mapping data to quantiles and isolating extreme values, QIM enhances the overall robustness of the dataset, ensuring more reliable predictive modeling. Subsequently, feature selection is carried out using PBB-ORF, a method tailored to identify and retain the most relevant features for attrition prediction. PBB-ORF leverages the power of partitioned binary bat optimization within an optimized random forest framework. This approach not only reduces dimensionality but also improves the model's interpretability and generalization performance. Through these preprocessing steps, the predictive model becomes more resilient to outliers and noise while focusing on the most discriminative features. This abstract lays the groundwork for a comprehensive study on improving employee attrition prediction accuracy and provides insights into the efficacy of advanced data preprocessing and feature selection techniques in HR analytics.