Knowledge-centric approaches in human resource management: leveraging clustering and deep learning

This research tackles contemporary human resource management challenges using advanced analytics methodologies. Initially, workforce dynamics are analysed through clustering to segment employees based on attributes. Among several algorithms evaluated, including K-means, agglomerative clustering, spectral clustering, and Gaussian mixture models, K-means proves most effective, with a Silhouette Score of 0.874156 and a Davies-Bouldin score of 1.285476. The study then predicts future skill requirements using deep learning models, focusing on the dense neural network. The dense NN emerges as the top predictive model, with the lowest mean squared error of 4478.58, the lowest mean absolute error of 47.56, and the highest R2 score of 0.94. Additionally, feature importance analysis highlights the dense NN's ability to capture intricate relationships, aiding HR practitioners in understanding key predictive factors. This research equips HR professionals with critical insights for proactive talent management and workforce planning.