In times past, when we wanted to know which team would win the World Cup, we had to turn to seers with crystal balls, use divination via tea leaves, or hope for Paul the Octopus to tell us what would happen.
But modern data science can provide a better alternative. As part of a team of statisticians, I helped train a machine learning algorithm to predict the most likely course of the tournament.
k9am!C@323:=:DE:4 7@C642DED 2?5 =@2565 5:46k^9am
kAm%96 2=8@C:E9> H6 3F:=E AC@4665D :? EH@ DE6AD]k^Am
kAmx? E96 7:CDE[ D@A9:DE:42E65 DE2E:DE:42= >@56=D 2?5 6IA6CE :?D:89E 7C@> 3@@<>2<6CD 2?5 EC2?D76C >2C<6ED 2C6 4@>3:?65 E@ 56E6C>:?6 E96 DEC6?8E9D @7 2== E62>D 2?5 E96:C A=2J6CD] x? E96 D64@?5 DE6A[ 2 >249:?6 =62C?:?8 2=8@C:E9> 564:56D 9@H E@ 36DE 4@>3:?6 E96 DEC6?8E9 6DE:>2E6D H:E9 @E96C :?7@C>2E:@? 23@FE E96 E62>D]k^Am
kAm%9:D AC@5F465 2 AC@323:=:DE:4 7@C642DE 7@C 6249 A@DD:3=6 >2E49 :? E96 E@FC?2>6?E] xE 42? 36 E9@F89E @7 2D 2 A2:C @7 =@2565 5:46i x?DE625 @7 92G:?8 E96 ?F>36CD ` E@ e H:E9 6BF2= AC@323:=:E:6D[ E96D6 =@2565 5:46 92G6 5:776C6?E AC@323:=:E:6D 7@C E96 ?F>36C @7 8@2=D 7@C 6:E96C E62>]k^Am
kAmu@C 6I2>A=6[ 244@C5:?8 E@ @FC 7@C642DE[ |6I:4@ 92D 2 5:6 C@==:?8 `]h 8@2=D @? 2G6C286 :? E96 @A6?:?8 >2E49[ H96C62D @AA@?6?E $@FE9 p7C:42 92D 2? 2G6C286 @7 @?=J _]f] qFE E9:D 5@6D ?@E >62? E92E |6I:4@ H:== DFC6=J H:?] #2E96C[ 2 H:? 7@C |6I:4@ :D E96 >@DE =:<6=J @FE4@>6 H:E9 edT AC@323:=:EJ] p 5C2H :D =6DD =:<6=J Wa`TX[ 2?5 2 H:? 7@C $@FE9 p7C:42 :D E96 =62DE =:<6=J @FE4@>6 W`cTX]k^Am
k9am‘'F6=G6 2 42D2[ 6= 7úE3@= GF6=G6 2 42D2P’k^9am
kAm&D:?8 5:776C6?E A2:CD @7 =@2565 5:46[ E96 C6DF=E @7 6249 >2E49 :? E96 (@C=5 rFA 42? 36 D:>F=2E65] (6 E@@< :?E@ 244@F?E E96 @77:4:2= E@FC?2>6?E 5C2H 2?5 2== uxup CF=6D[ :?4=F5:?8 E96 A@DD:3:=:EJ @7 @G6CE:>6 2?5 A6?2=EJ D9@@E@FED] (6 C2? E96 D:>F=2E:@? `__[___ E:>6D E@ 56E6C>:?6 E96 E@FC?2>6?E’D >@DE =:<6=J 4@FCD6]k^Am
kAm%96 C6DF=ED D9@H E92E $A2:? :D E96 72G@C:E6 7@C E96 E:E=6 H:E9 2 H:??:?8 AC@323:=:EJ @7 `c]dT[ 4=@D6=J 7@==@H65 3J t?8=2?5 2?5 uC2?46[ 6249 2E `a]cT[ 2?5 v6C>2?J 2E ``]aT]k^Am
kAmk:>8 DC4lQ9EEADi^^52E2HC2AA6C]5H45?]?6E^;~p3)^7F==]A?8Q ^mk^Am
kAmsF6 E@ E96 6IA2?565 E@FC?2>6?E – E9:D (@C=5 rFA 92D cg E62>D 2?5 7:G6 C@F?5D :? E96 <?@4<@FE DE286 – E9:D 8C@FA @7 72G@C:E6D :D E:89E=J A24<65] !@CEF82= 2?5 pC86?E:?2 2=D@ 92G6 8@@5 492?46D E@ H:? E96 E:E=6[ 2E g]hT 2?5 g]aT[ C6DA64E:G6=J]k^Am
kAmu@C :ED A2CE[ E96 &?:E65 $E2E6D 92D 2 8@@5 492?46 @7 C6249:?8 E96 #@F?5 @7 bai fgT] %9:D :D E96 9:896DE :? E96:C 8C@FA[ H9:49 92D E9C66 @E96C E62>D] x? E96 <?@4<@FE DE286[ 9@H6G6C[ H96? 6G6CJ >2E49 :D 5@ @C 5:6[ E96 AC@323:=:E:6D @7 E96 &]$] E62> “DFCG:G:?8” 8@ 5@H? C6=2E:G6=J BF:4<=J] %96 AC@323:=:EJ 7@C 2 9@>6 G:4E@CJ :? E96 7:?2= 2E |6E{:76 $E25:F> :? }6H y6CD6J @? yF=J `h :D `T]k^Am
k9amp 566A6C A66< :?E@ E96 6?8:?6 C@@>k^9am
kAm~FC >249:?6 =62C?:?8 2=8@C:E9> 2?5 DF3D6BF6?E D:>F=2E:@?D 2C6 7F6=65 3J 52E2[ 6IA6CE <?@H=6586 2?5 DE2E:DE:42= >@56=D]k^Am
kAmu:CDE[ 2== ?2E:@?2= >2E496D @G6C E96 A2DE 6:89E J62CD 2C6 E96 32D:D 7@C 2 “C6EC@DA64E:G6” 6DE:>2E6 @7 E96 E62>D’ DEC6?8E9D] $64@?5[ 2 “AC@DA64E:G6” DEC6?8E9 6DE:>2E6 :D @3E2:?65 7C@> BF@E65 @55D @7 G2C:@FD :?E6C?2E:@?2= 3@@<>2<6CD[ C67=64E:?8 E96:C 6IA6CE @A:?:@?D 23@FE E96 FA4@>:?8 E@FC?2>6?E]k^Am
kAm%9:C5[ C2E:?8D @7 E96 :?5:G:5F2= A=2J6CD 2C6 AC@5F465 32D65 @? E96:C 4@?EC:3FE:@?D E@ 8@2=D 2E E96 4=F3 2?5 ?2E:@?2= =6G6=D] p?5 7:?2==J[ E96 4FCC6?E BF2=:EJ 2?5 7FEFC6 A@E6?E:2= @7 E96 A=2J6CD :D C67=64E65 :? E96:C 6IA64E65 >2C<6E G2=F6D] %96D6 2C6 2G2:=23=6 7C@> E96 k2 9C67lQ9EEADi^^HHH]EC2?D76C>2C<E]4@>^Qm%C2?D76C>2C<E H63D:E6k^2m E92E FD6D k2 9C67lQ9EEADi^^HHH]EC2?D76C>2C<E]4@>^?2G:82E:@?^>H567:?:E:@?Qm2 H:D5@>\@7\E96 4C@H5 2AAC@249k^2m E@ 6DE:>2E6 E96 F?<?@H? C62=\>2C<6E G2=F6D]k^Am
kAm%96D6 7@FC G2C:23=6D 2C6 4@>3:?65 H:E9 2 3C@25 C2?86 @7 7FCE96C C6=6G2?E :?AFED C67=64E:?8 E96 4FCC6?E DE2E6D @7 E96 5:776C6?E E62>D 2?5 E96 4@F?EC:6D E96J 4@>6 7C@>] %9:D :?4=F56D E62>\DA64:7:4 56E2:=D[ DF49 2D E96:C k2 9C67lQ9EEADi^^:?D:56]7:72]4@>^7:72\H@C=5\C2?<:?8^>6?Qmuxup C2?<k^2m 2?5 E96 ?F>36C @7 A=2J6CD :? E96 D6>:7:?2=D @7 E9:D J62C’D r92>A:@?D {628F6] (6 2=D@ 724E@C65 :? 4@F?ECJ\DA64:7:4 D@4:@64@?@>:4 724E@CD[ DF49 2D vs! A6C 42A:E2]k^Am
kAm%@ 56E6C>:?6 :7 2?5 9@H E96D6 762EFC6D 2C6 C6=6G2?E 7@C E96 24EF2= C6DF=ED :? 2 (@C=5 rFA[ 2 >249:?6 =62C?:?8 2=8@C:E9> H2D FD65]k^Am
kAmw6C6[ 2 D@\42==65 k2 9C67lQ9EEADi^^HHH]866<D7@C866<D]@C8^>249:?6\=62C?:?8^C2?5@>\7@C6DE\2=8@C:E9>\:?\>249:?6\=62C?:?8^QmC2?5@> 7@C6DEk^2m :D EC2:?65[ 4@?D:DE:?8 @7 =@ED @7 564:D:@? EC66D 42AEFC:?8 D=:89E=J 5:776C6?E DF3D6ED @7 E96 52E2] %96 2=8@C:E9> 92D 366? EC2:?65 @? 2== >2E496D A=2J65 2E E96 >2;@C D@446C E@FC?2>6?ED D:?46 (@C=5 rFA a__e] xE E9FD =:?<D 2 E62>’D DEC6?8E9[ >2C<6E G2=F6 2?5 @E96C 724E@CD E@ E96 ?F>36C @7 8@2=D D4@C65 :? >2E496D 2E (@C=5 rFAD] %9:D :D E96 :?7@C>2E:@? E92E =@25D E96 5:46 7@C @FC D:>F=2E:@?D]k^Am
k9amu:?5 @FE >@C6k^9am
kAm%9:D :D ?@E E96 7:CDE E:>6 E92E @FC E62> 4@>AC:D:?8 k2 9C67lQ9EEADi^^35]DE2E:DE:<]EF\5@CE>F?5]56^AC@76DDFC^2C36:ED8CFAA6^AC@7\5C\2?5C62D\8C@==^Qmp?5C62D vC@==k^2m 2?5 k2 9C67lQ9EEADi^^C@FG6?>]8:E9F3]:@^Qm#@FG6? |:496=Dk^2m 2?5 4@==628F6D 2E %& s@CE>F?5 &?:G6CD:EJ :? v6C>2?J[ k2 9C67lQ9EEADi^^9@>6]9:>@=56]?@^9G2EEF>^Qm{2CD |28?FD wG2EEF>k^2m 2E }@CH2J’D |@=56 &?:G6CD:EJ r@==686[ k2 9C67lQ9EEADi^^HHH]D8]EF>]56^6A:56>:@=@8:6^E62>^D492F36C86C^QmvF?E96C $492F36C86Ck^2m 2E %& |F?:49 2?5 x 92G6 4@==23@C2E65 E@ 7@C642DE 2 (@C=5 rFA]k^Am
kAmx? E96 a_`h (@>6?’D (@C=5 rFA H6 4@CC64E=J AC65:4E65 E96 &]$] 2D E96 H:??6C] x? E96 a_ab (@>6?’D (@C=5 rFA 2?5 E96 a_aa >6?’D (@C=5 rFA[ E96 H:??6CD – $A2:? 2?5 pC86?E:?2[ C6DA64E:G6=J – H6C6 ?@E @FC 72G@C:E6D[ 2=E9@F89 H6 5:5 AC65:4E E96> E@ 36 D6C:@FD 4@?E6?56CD]k^Am
kAm%96 3@EE@> =:?6 :D 7@C642DED 2C6 23@FE AC@323:=:E:6D] ~FC AC@8C2> H:== ?@E AC65:4E E96 H:??6C H:E9 `__T 46CE2:?EJ – 3FE :E >:89E 5@ 36EE6C E92? 2? 6:89E\=:>365 >@==FD<]k^Am
kAmk:>8 DC4lQ9EEADi^^52E2HC2AA6C]5H45?]?6E^Et8<c^7F==]A?8Q ^mk^Am
kAmk6>m%9:D 2CE:4=6 :D C6AF3=:D965 7C@> k2 9C67lQ9EEAi^^E964@?G6CD2E:@?]4@>^FDnFE>0D@FC46l{66U2>AjFE>0>65:F>lC6=2E65\=:?<U2>AjFE>042>A2:8?lC6=2E65\=:?<\_U2>AjFE>04@?E6?El2CE:4=6\agceahQm%96 r@?G6CD2E:@?k^2m[ 2 ?@?AC@7:E[ :?56A6?56?E ?6HD @C82?:K2E:@? 3C:?8:?8 J@F 724ED 2?5 ECFDEH@CE9J 2?2=JD:D E@ 96=A J@F >2<6 D6?D6 @7 @FC 4@>A=6I H@C=5] xE H2D HC:EE6? 3Ji k2 9C67lQ9EEADi^^E964@?G6CD2E:@?]4@>^AC@7:=6D^249:>\K6:=6:D\af_a`bcnFE>0D@FC46l{66U2>AjFE>0>65:F>lC6=2E65\=:?<U2>AjFE>042>A2:8?l3J=:?6\=:?<U2>AjFE>04@?E6?El2CE:4=6\agceahQmp49:> +6:=6:Dk^2m[ k6>mk2 9C67lQ9EEADi^^E964@?G6CD2E:@?]4@>^:?DE:EFE:@?D^F?:G6CD:EJ\@7\:??D3CF4<\c`hgnFE>0D@FC46l{66U2>AjFE>0>65:F>lC6=2E65\=:?<U2>AjFE>042>A2:8?l3J=:?6\=:?<U2>AjFE>04@?E6?El2CE:4=6\agceahQm&?:G6CD:EJ @7 x??D3CF4<k^2mk^6>mk^6>mk^Am
kAmkDEC@?8m#625 >@C6ik^DEC@?8mk^Am
kF=mk=:mk2 9C67lQ9EEAi^^E964@?G6CD2E:@?]4@>^D@446C\:D\2\7:?6\E6C>\7@C\E96\362FE:7F=\82>6\5@?E\=6E\2?J\7@@E32==\D?@3\@C\AC6D:56?E\E6==\J@F\@E96CH:D6\E9:D\H@C=5\4FA\ag_ffhnFE>0D@FC46l{66U2>AjFE>0>65:F>lC6=2E65\=:?<U2>AjFE>042>A2:8?lC6=2E65\=:?<_U2>AjFE>04@?E6?El2CE:4=6\agceahQm‘$@446C’ :D 2 7:?6 E6C> 7@C E96 362FE:7F= 82>6 – 5@?’E =6E 2?J ‘7@@E32==’ D?@3 @C AC6D:56?E E6== J@F @E96CH:D6 E9:D (@C=5 rFAk^2mk^=:mk=:mk2 9C67lQ9EEAi^^E964@?G6CD2E:@?]4@>^D@2C:?8\E:4<6E\AC:46D\4@F=5\96=A\7:72\AF==\:?\`d3\E9:D\H@C=5\4FA\4J4=6\H96C6\5@6D\E96\>@?6J\4@>6\7C@>\H96C6\5@6D\:E\8@\aff`agnFE>0D@FC46l{66U2>AjFE>0>65:F>lC6=2E65\=:?<U2>AjFE>042>A2:8?lC6=2E65\=:?<`U2>AjFE>04@?E6?El2CE:4=6\agceahQm$@2C:?8 E:4<6E AC:46D 4@F=5 96=A uxup AF== :? q E9:D (@C=5 rFA 4J4=6 — H96C6 5@6D E96 >@?6J 4@>6 7C@>[ H96C6 5@6D :E 8@nk^2mk^=:mk=:mk2 9C67lQ9EEAi^^E964@?G6CD2E:@?]4@>^A9:=256=A9:2D\a_ae\DA@CED\42=6?52C\:D\A24<65\3FE\72?D\2C6\36:?8\AC:465\@FE\agaf__nFE>0D@FC46l{66U2>AjFE>0>65:F>lC6=2E65\=:?<U2>AjFE>042>A2:8?lC6=2E65\=:?<aU2>AjFE>04@?E6?El2CE:4=6\agceahQm!9:=256=A9:2’D a_ae DA@CED 42=6?52C :D A24<65 – 3FE 72?D 2C6 36:?8 AC:465 @FEk^2mk^=:mk^F=m
kA 4=2DDlQ7:?6\AC:?EQmk6>mkDA2?mp49:> +6:=6:D 5@6D ?@E H@C< 7@C[ 4@?DF=E[ @H? D92C6D :? @C C646:G6 7F?5:?8 7C@> 2?J 4@>A2?J @C @C82?:K2E:@? E92E H@F=5 36?67:E 7C@> E9:D 2CE:4=6[ 2?5 92D 5:D4=@D65 ?@ C6=6G2?E 277:=:2E:@?D 36J@?5 E96:C 24256>:4 2AA@:?E>6?E]k^DA2?mk^6>mk^Am
kAmk:>8 DC4lQ9EEADi^^4@F?E6C]E964@?G6CD2E:@?]4@>^4@?E6?E^agceah^4@F?E]8:7n5:DEC:3FE@Cl=66Q DEJ=6lQ96:89Ei`AIjH:5E9i`AIjQ 2=ElQQ ^mk^Am

