This study was conducted to explore major variables explaining career attitudes of multicultural adolescents to discuss the implications for career development of multicultural adolescents, who have difficulty in career development, as interest in personalized career education increases. To do this, random forest, one of the machine learning techniques, was applied to the 9th wave data (11th graders) from MAPS (the Multicultural Adolescents Panel Study). As a result, 20 major variables were derived by using the SHAP (Shapley Additive Explanation) value. The contribution of predictions to career attitudes was high in the order of career barrier (lack of self-understanding), achievement motivation, multicultural acceptance, academic adaptation, and bicultural acceptance attitude. In addition, national identity, interest in foreign parents' countries, career barrier (lack of career and job information), planning after high school graduation, overall satisfaction with career activities in school, and self-esteem were derived as major variables. Moreover, variables related to social relationships were derived from among environmental factors, and variables representing relationships with parents (such as parents' frequency of conversation and parental support) and teachers and friendship were the major variables predicting career attitude. Accordingly, this study examined the tendency of career attitudes according to the response level of the major variables and discussed ways to support personalized career education for career development of multicultural adolescents based on the results.