Using Non-Negative Matrix Factorization to Identify the Factors Affecting the Performance of Bank Employees in Australia

Identifying, training, and hiring new workforce is one of the greatest challenges for any organization. Hence, managers of organizations tend to keep their professional workforce for a long time. This issue is more critical for the human resource managers of banks because it is difficult to attract expert personnel who are familiar with the financial and credit field. Therefore, to clarify the challenges in work environment, we look for a method to investigate this issue by exploring and analysis of opinions of banking industry employees. The method used in this research is the non-negative matrix factorization method. To perform the model optimally, two feature engineering techniques called TF-IDF and Bag of Words with two types of functions, Frobenius norm, and Kullback-Leibler divergence have been implemented. To evaluate the model, we used the silhouette clustering index and the cohesion criterion. Finally, the BoW method with the KL-Divergence function is the optimal model with the silhouette criterion with a value of 0.36 and the cohesion criterion with a value of 0.8945. After modeling, 12 major topics that are of interest to bank employees have been identified. After analyzing results obtained from NMF modeling "management", "career opportunity", "people and organization" are the most dominant topics that appear in employee opinions with the frequency of 12.44%, 9.78%, and 9.46%, respectively. On the other hand, "customer service", "role-play on teamwork" and "organizational culture" have the least impact on how employees evaluate the work environment with the frequency of 6.22%, 6.53%, and 6.89%. The results of this research can provide better insight for human resource managers in Australia for evaluating banking industry employees' performance in the workplace.