With the advent of information-based society, career information has become increasingly important to students. The main purpose of this study was to explore variables to predict high-school students’ career information needs and consequently contribute to improving career information programs. In particular, this study employed group Mnet, a penalized regression method (machine learning), to analyze students, their parents, homeroom teachers, career counseling teachers, and school administrators data of Career Education Status Survey (2018). After 100 times of modeling with random data splitting, a total of 25 student variables and 5 parent variables were selected as important. Selected student variables included gender, school satisfaction, educational aspirations, learning motivation, career planning after high-school graduation, career education activities, career awareness, and risk-taking disposition. Parent variables included age, career guidance needs, and satisfaction on school support. Based on the results of the study, suggestions for career information programs are discussed.