中文摘要：

我們都知道：「青少年是國家未來的希望和根本。」但是在青少年的成長階段過程中，會面臨到生理、心理、家庭、社會、學校等多重因素的影響，這些因素都有可能造成少年的行為價值觀偏差，進而惡化形成犯罪行為。犯罪學者常言：「少年犯罪起因於家庭，顯現於學校，為害於社會。」由此可知，我們不斷地在探究摸索為何會產生少年犯罪現象的原因背後，其實反映的正是一個複雜的社會問題，而「今日之少年犯很可能成為明日之成年犯」，在少年心智、價值觀尚未成熟的這個重要關鍵階段，如何找出潛藏的少年高風險族群，並有效地運用政府相關社政資源來予以協助關懷、甚至是輔導矯正，這也是世界各國一直以來執行少年犯罪預防工作的重點。
本研究希望透過彙整新北市政府警察局既有少年不良行為、曝險行為及犯罪行為資料，結合相關犯罪學理論，利用機器學習的方式，從中找出一些影響少年犯罪的關鍵成因，並建立一套分析模型用以找尋潛藏的少年犯罪高風險族群，希望未來能夠有效作為外勤員警執行犯罪預防偵查，以及學校、少年輔導團體關懷輔導之參考依據，並藉由引進多元資源，整合教育、社政、衛生、法務、司法等相關專業體系資源，強化政府社會安全網，進而達成降低社會整體犯罪率之目標。

英文摘要：

As we all know, juveniles are the hope and fundamental of country’s future. But juveniles will be facing impacts from multiple factors including physiology, psychology, family, society, school during the growth phase. All these factors may cause the deviation of juvenile behavioral values, a criminal act even worse. Criminologists always said, juvenile delinquency formed in family, found in school, and costed in society. Juvenile delinquency actually reflect a complicated society problem, that’s why we keep explore and grope what may cause it. There is a saying, juvenile offender probably become criminal in the future. So it’s crucial to find hidden high-risk teenagers and provide resources to care, counsel, and even correction in the important phase when teenagers are not mature enough in mind and values. This is the first priority in juvenile delinquency prevention as well all around the world.
This study made to consolidate data of delinquent conduct, perpetration, and criminal act of New Taipei City Police Department, combine with relevant criminological theory, using machine learning to determine key factors of causing juvenile delinquency. Therefore, we could build an analyzing model to search for hidden high-risk groups as a reference for field officers to carry out crime prevention, school and NGO to conduct a counseling. In addition, to introduce and integrate resources from education, social administration, health, and judicial to strengthen social security network in order to reduce overall crime rate.

文章連結：

https://ndltd.ncl.edu.tw/cgi-bin/gs32/gsweb.cgi/ccd=W1MOi9/record?r1=4&h1=0#XXX

資料來源：

臺灣博碩士論文知識加值系統