This　paper　used　machine　learning　to　model　the　prediction　of　creep　and　the　analysis　of　characteristic　factors　for　concrete　containing　supplementary　cementitious　materials　(SCM).　First,　a　creep　database　covering　thirteen　input　parameters　and　one　output　parameter　was　developed.　Then,　based　on　this　database,　Linear　Regression　(LR),　Support　Vector　Regression　(SVR),　Random　Forest　(RF)　and　Extreme　Gradient　Boosting　(XGB)　models　were　used　to　establish　creep　compliance　models.　Finally,　the　sensitivity　analysis　of　each　input　parameter　based　on　the　shapley　additive　explanation　(SHAP)　model　was　performed　to　investigate　the　influence　of　each　parameter　on　the　prediction　results　of　creep.　The　results　show　that　the　extreme　gradient　boosting　(XGB)　model　can　predict　the　creep　compliance　of　SCM　concrete　very　well.　The　creep　compliance　is　the　result　of　the　combined　effect　of　numerous　positively　and　negatively　correlated　characteristic　parameters.　Among　the　many　characteristic　pa-rameters,　the　stress-to-strength　ratio　is　the　most　important　factor　affecting　the　prediction　of　creep.　The　creep　compliance　of　concrete　is　positively　correlated　with　cement　content,　while　it　is　negatively　correlated　with　silica　fume　(SF)　content,　and　shows　a　negative　and　then　positive　change　with　slag　and　fly　ash　(FA)　content.　The　creep　tendency　is　smaller　when　the　slag　content　is　about　25%　or　the　FA　content　is　about　20%.　Meanwhile,　the　interaction　interval　between　different　SCM　admixtures　and　cement　content　differed　in　concrete　creep　prediction.