Reducing　carbon　emissions　is　an　ongoing　goal　of　the　whole　world　and　its　achievement　requires　an　outstanding　approach　to　accurately　predict　future　carbon　emissions　and　explore　the　factors　driving　carbon　emissions.　Hence,　this　study　proposes　a　driving　factor　decomposition-based　data-driven　rule-base　(DFD-DDRB)　approach　for　the　aim　of　analyzing　carbon　emission　reduction　pathway　from　predictive　perspective,　where　the　approach　includes　three　processes:　1)　generating　a　rule-base　from　historical　carbon　emission　data;　2)　predicting　multi-scenario　carbon　emissions　using　the　rule-base;　3)　providing　predictive　analytics　for　future　carbon　emission　reduction.　In　empirical　study,　the　China＇s　provincial　data　from　2004　to　2021　are　used　to　justify　the　applicability　of　the　proposed　approach.　The　experimental　findings　not　only　show　that　the　approach　can　accurately　predict　multi-scenario　carbon　emissions　until　2035　and　reveal　the　factors　driving　carbon　emissions,　but　also　provide　three　implications　for　reducing　China＇s　carbon　emissions:　1)　resource　endowment　should　be　considered　to　establish　carbon　emission　management　policies　of　30　Chinese　provinces;　2)　economic　development　effect　can　be　regarded　as　the　main　factor　driving　China＇s　future　carbon　emissions;　3)　optimizing　energy　structure　and　consumption　is　much　important　for　reducing　China＇s　provincial　carbon　emissions.　Beside　the　work　in　China,　the　DFD-DDRB　approach　can　be　also　used　as　the　generic　analytical　framework　served　for　some　developed　economies　and　other　carbon-emitting　countries.　©　2025　Elsevier　Ltd