Education　is　something　that　every　country　values,　and　education　data　is　a　very　important　resource　for　the　country.　With　the　increase　in　the　proportion　of　education　in　the　country,　the　size　of　the　student　body　is　getting　bigger　and　bigger.　Student　performance　is　directly　related　to　the　core　of　the　entire　education.　By　analyzing　the　student＇s　information　and　predicting　the　student＇s　future　performance,　this　prediction　does　not　only　mean　the　improvement　of　the　student＇s　grades,　but　also　can　summarize　the　methods　that　can　effectively　help　the　student　avoid　the　above　situation.　In　this　study,　some　of　the　various　algorithms　in　machine　learning　will　be　analyzed,　such　as　linear　regression,　ridge　regression,　and　decision　trees,　have　higher　accuracy　and　are　more　suitable　for　inferring　student　achievement.　In　addition,　this　work　analyzes　whether　this　model　can　improve　the　accuracy　of　machine　learning　for　student　performance　prediction　by　pre-processing　the　student　information　dataset　to　find　several　label　items　that　have　a　greater　impact　on　student　performance,　and　focus　on　using　these　label　items　for　machine　learning.