Predicting Adolescent Deviant Behaviors through Data Mining Techniques

Adolescence is the time during which people develop and form their crucial values, personality traits, and beliefs. Hence, as deviant behaviors occur during adolescence, it is important to guide adolescents away from such behaviors and back to normal behaviors. Moreover, although there are various kinds of deviant behavior, most of them would either directly or indirectly affect youths’ academic learning progresses. Therefore, many researchers have endeavored to explore the issues of juvenile delinquency. In this study, we focus on providing methods that could assist counseling officers in discovering symptoms and pre-symptoms of youth delinquency. The study proposes a framework for mining associations from “memo-type” records without supervision of senior experts. In addition, we apply this approach to retrieve associations among deviant behaviors from the counseling records stored in databases. The contributions to adolescent counseling are as follows: (1) A keyword tree of deviant behaviors is constructed and verified; (2) the general and time sequence associations of deviant behaviors are mined from memo-type records and also verified. Finally, (3) an information system is recommended to help junior counselors performing counseling works. Consequently, without experienced consultants, the proposed framework could discover valuable knowledge from consulting records effectively and many efforts are therefore saved.