Talent Selection and Cluster Analysis using Machine Learning

The purposes of this research were to construct a model for selecting talented employees and clustering talented employees. The data such as employee age (years), year of experience (years), and performance results of 13,933 the provincial electricity authority employees in 2021 were employed for models. They were divided into 2 sets: a training set (9,753 employees) for constructing model and a test set (4,180 employees) for evaluating the model performance. The 7,754 talented employees were clustered using K-Means and validated by Silhouette statistic.

It has been found that the CART model was the best model with the highest accuracy, followed by SVM model, logistic regression model, and the naïve bayes model, respectively. The factors that affect the selection of talented employees were employees' age, performance, and year of experience, respectively. In addition, there were 3 clustering groups for talented employees, the excellent performance (51.68%), the moderate performance (32.99%) and the low performance (15.33%).