Through　the　application　and　expansion　of　expressway　ETC　gantry　transaction　data,　we　propose　a　short-term　traffic　flow　forecasting　of　expressway　based　on　the　Kalman　Filtering　(KF)　and　Random　Forest　(RF)　model,　which　not　only　takes　into　account　the　basic　external　features　and　periodic　features　but　also　considers　the　spatio-temporal　correlation　relationship　in　the　road　section,　so　as　to　construct　the　spatial　correlation　features　and　temporal　correlation　features.　In　this　paper,　we　use　the　ETC　gantry　transaction　data　of　Fuzhou–Xiamen　section　of　the　expressway　to　forecast　and　verify　in　Fujian　Province,　China,　the　final　results　show　that:　When　the　rolling　window　is　20 min,　compared　with　the　results　before　and　after　Kalman　Filtering　algorithm　processing　traffic　flow　data,　the　performance　indicators　is　greatly　improved,　which　verifies　the　positive　effect　of　Kalman　Filtering　algorithm;　it　is　also　verified　that　the　constructed　features　have　a　great　influence　on　traffic　flow　forecasting　and　play　a　positive　role　in　improving　forecasting　accuracy;　and　it　is　also　verified　that　the　RF　model　has　better　forecasting　effect　than　the　baseline　models.　©　2023,　The　Author(s),　under　exclusive　license　to　Springer　Nature　Singapore　Pte　Ltd.