KNN　is　an　important　part　of　classification　algorithms　in　machine　learning,　which　has　been　extensively　employed　in　various　fields.　In　this　research,　we　proposed　a　Lp　Norm-Based　Local　Means　k-Nearest　Neighbor　Classification　with　Feature　Reduction.　We　extend　the　original　Euclidean　distance　formula,　and　obtain　the　distance　calculation　formula　based　on　P-value.　Then　classify　and　predict　the　data　set　after　dimension　reduction.　In　the　experiment,　we　verified　the　effectiveness　and　superiority　of　our　method　by　setting　different　k　values　in　KNN,　LMKNN　and　the　classification　accuracy　obtained　by　our　methodology　(PLMKNN).　　©　2023　IEEE.