The purpose of this study is to classify latent profiles of job satisfaction of employees with turnover experience, and to explore the factors affecting latent profiles by using a machine learning approach. To do this, Youth Panel (YP) data from the 1st to 11th waves were used. The results of this study include: first, according to latent profile analysis, the levels of job satisfaction were classified into four latent profiles, namely, ‘low job satisfaction,’ ‘middle job satisfaction,’ ‘high job satisfaction,’ and ‘highest job satisfaction.’ Second, the random forest method had a higher classification accuracy than the decision tree method. The random forest analysis revealed that the main factors associated with the classified latent profiles were wage, given annual leave, household earned income, average savings, age, stress, health status, job-major congruence, required level of education, and turnover frequency. Based on these results, this study suggested ways to improve the job satisfaction of employees.