The　traditional　economic　development　mode　can　no　longer　meet　the　increasingly　stringent　demands　for　innovation　and　environmental　sustainability.　Consequently,　green　innovation　has　emerged　as　a　pivotal　development　trend.　Accurate　measurement　of　green　innovation　efficiency　is　crucial　in　this　context.　This　study　employs　the　SBM-DDF-GML　model　to　evaluate　the　green　innovation　efficiency　of　30　Chinese　provinces　from　2000　to　2020,　incorporating　enhanced　indicators　for　both　expected　and　unexpected　outputs.　Additionally,　the　K-means　algorithm,　a　machine　learning　technique,　was　utilized　to　cluster　the　comprehensive　development　factors　of　these　provinces,　enabling　an　analysis　of　their　spatiotemporal　heterogeneity.　The　findings　indicate　that　the　improved　model　enhances　the　precision　of　regional　green　innovation　efficiency　rankings,　providing　a　more　accurate　reflection　of　actual　regional　changes.　Furthermore,　compared　to　traditional　regional　classification　methods,　the　K-means　clustering　based　on　comprehensive　regional　development　factors　exhibited　greater　inter-group　differences,　aligning　more　closely　with　the　heterogeneity　analysis　of　regional　green　innovation　efficiency.　The　spatiotemporal　heterogeneity　analysis　of　the　new　groupings　revealed　that　the　evolution　of　green　innovation　efficiency　is　predominantly　influenced　by　advancements　in　green　innovation　technology.