Based　on　machine-learning　(ML)　and　analytical　methods,　a　hybrid　method　is　developed　herein　to　predict　the　ground-displacement　field　(GDF)　caused　by　tunneling.　The　extreme　learning　machine　(ELM),　as　a　single　hidden　layer　feedforward　neural　network,　is　used　as　an　ML　model　to　predict　maximum　settlement　s(max)　of　the　ground　surface.　The　particle　swarm　optimization　(PSO)　algorithm　is　applied　to　optimize　the　parameters　for　the　ELM　method,　namely,　weight　and　bias　values　from　the　input　layer　to　the　hidden　layer.　The　mean　square　error　of　the　k-fold　cross　validation　sets　is　considered　the　fitness　function　of　the　PSO　algorithm.　For　38　data　samples　from　published　papers,　30　samples　are　used　as　the　training　set,　and　8　samples　are　used　as　the　test　set.　For　the　test　samples,　the　error　of　five　samples　ranges　between　-5　and　5　mm.　The　error　of　only　one　sample　is　slightly　greater　than　10　mm.　The　proposed　PSO-ELM　method　demonstrates　good　prediction　performance　of　s(max).　A　deformation　parameter　of　the　nonuniform　displacement　mode　for　the　tunnel　cross-section　is　calibrated　based　on　predicted　s(max).　When　the　determined　nonuniform　displacement　mode　is　used　as　the　boundary　condition　of　the　tunnel　cross-section,　the　GDF　of　a　shallow　circular　tunnel　is　analytically　predicted　based　on　the　complex-variable　method　prior　to　tunnel　excavation.　For　a　specific　engineering　case,　i.e.,　the　Heathrow　Express　tunnel,　the　proposed　PSO-ELM-analytical　method　can　well　predict　the　surface-settlement　trough　curve,　horizontal　displacements　at　different　depths,　and　vertical　displacements　above　the　tunnel.