Machine Learning Integration of Herzberg’s Theory using C4.5 Algorithm

Abstract

The emergence of the global economy, particularly to the ASEAN Community in 2015, marked a significant milestone on its path towards becoming a highly competitive region in terms of the labor force integrated into the global economy. Human Resource Management Society revealed that businesses would spend the equivalent of six to nine times the monthly wage of an employee to locate and train a replacement. The Philippines is recognized to have a high aptitude skilled workforce and top talent. The I.T. industry became diversified, where sophisticated skillset is part of job hiring requirements. Retention of employees became a critical success factor of an organization. In the 1950s, Fredrick Herzberg, a psychiatrist, carried out a study of employee satisfaction. Herzberg proved in his experimentation test that job satisfaction and dissatisfaction are related to workplace environment factors. Some situations make an employee engaged and discouraged, which affects the attitude towards the job. The Machine Learning, as part of Data Mining classification tasks towards a knowledge acquisition process. It is categorized and carried out through the analysis. This method is intending to use talent databases to conduct the process of acquiring talent expertise and retention pattern in Human Resource (H.R.). The decision tree family C4.5 classifier algorithm suggests an appropriate dataset classifier. The classification method performed is using the management of talent for the identification or estimation of retention of employees. The decision tree induction adopts a top-down approach, which begins with a tuple training set and related class labels. As the tree constructs, the training set is repetitively dividing into smaller subsets. The result of this study affirms Herzberg's theory of satisfaction through a mathematical technique C4.5 Algorithm. The employment retention model using the machine learning technique identifies the dataset properties that can avoid business disruptions in multiple ways. It is measuring the robustness of the datasets using WEKA, giving a 99.06% data performance.