坂本勇人選手はいつ通算3,000安打を達成するか? AIに聞いてみました / Hayato Sakamoto Performance Prediction Using Feature Engineering with Machine Learning and Python

Slide 1

Slide 1 text

ӫޫͷഎ൪߸6⃣ ࡔຊ༐ਓ3,000ຊ҆ଧه೦LT Shinichi Nakagawa(@shinyorke) Sports Analyst Meetup #9 2020/12/13

Slide 2

Slide 2 text

ʁʁʁʮ༐ਓ͸·ͩ2,000ຊ҆ଧ΍Ζʯ

Slide 3

Slide 3 text

ͦͷͱ͓ΓͰ͍͟͝·͢, ࣦྱ͠·ͨ͠

Slide 4

Slide 4 text

ࡔຊ༐ਓ͸͍ͭ௨ࢉ3,000ຊ҆ଧΛ ୡ੒͢Δ͔AIʹฉ͍ͯΈ·ͨ͠ Shinichi Nakagawa(@shinyorke) Sports Analyst Meetup #9 2020/12/13

Slide 5

Slide 5 text

ຊ೔ͷςʔϚ • ࡔຊ༐ਓ͕͍ͭ͝Ζ௨ࢉ3,000ຊ҆ଧΛୡ੒͢Δ͔༧૝͢Δ • ਅ໘໨ͳ࿩, ੒੷༧ଌ͸Ͳ͜·ͰͰ͖Δ͔ࢼͯ͠ΈΔ • ʮӫޫͷഎ൪߸6⃣ࡔຊ༐ਓ3,000ຊ҆ଧ΁ͷಓʯ͕   Կ೥ޙʹ์ө͞ΕΔ͔Θ͔Δ΄͏͕͍͍ΑͶʢద౰ʣ

Slide 6

Slide 6 text

Who am I ?ʢ͓લ୭Αʣ • Shinichi Nakagawaʢத઒ ৳Ұʣ • େ఍ͷSNSͰʮshinyorkeʢ͠ΜΑʔ͘ʣʯͱ໊৐͍ͬͯ·͢ • JX Press Corporation Senior Engineer   ʢJX௨৴ࣾ γχΞɾΤϯδχΞʣ • Baseball Engineer, Data Scientist   ʢ໺ੜͷ໺ٿΤϯδχΞɾσʔλαΠΤϯςΟετʣ • Ҏલ͸͓࢓ࣄͰ໺ٿΤϯδχΞʮͩͬͨʯਓ

Slide 7

Slide 7 text

ʲCMʳαʔόʔαΠυΠϯλʔϯืूͯ͠·͢ https://www.wantedly.com/projects/543767 ※ֶੜ͞ΜݶఆͰ͢&ผʹεϙʔπͷ࢓ࣄͬͯ༁͡Όͳ͍Ͱ͢

Slide 8

Slide 8 text

26.4ඵͰৼΓฦΔ2020೥ͷϓϩ໺ٿ • ιϑτόϯΫϗʔΫε೔ຊҰʢ4࿈೼ʣ • όϯςϦϯυʔϜφΰϠ&౦ژυʔϜͷ਌ձࣾ(ry • ࡔຊ༐ਓʢڊਓʣ, ӈଧऀͱͯ͠࠷೥গͰ2,000ຊ҆ଧୡ੒ ͦͷଞʹ΋͍ͬͺ͍͋Δ͚ͲׂѪʢద౰ʣ

Slide 9

Slide 9 text

ࡔຊ༐ਓબखͳΒ3,000ຊ҆ଧ༨༟Ͱ͸ • 31ࡀ10ϲ݄Ͱͷୡ੒͸ӈଧऀ࠷଎ • গͳ͘ͱ΋͋ͱ4, 5೥͸ݱ໾͢ΔͰ͠ΐ   γϣʔτͰݩؾʹΠέͯ·͢͠. • ͡Ό͍͋ͭࠒ3,000ຊ҆ଧ΍Δͷ͞? ͜Εͬͯաڈͷσʔλ͔Β͏·͍۩߹ʹ΍Ε͹༧ଌՄೳͰ͸? https://www.nikkansports.com/baseball/news/202011080000831.html

Slide 10

Slide 10 text

ͱ͍͏Θ͚Ͱ༧ଌϞσϧΛ࡞Γ·ͨ͠. ࠓճ͸PyCon JP 2020Ͱ΍ͬͨ΍ͭΛݩʹͪΐͬͱΞϨϯδͯ͠࡞Γ·ͨ͠. https://shinyorke.hatenablog.com/entry/baseball-and-ml-with-python

Slide 11

Slide 11 text

ࠓճͷΞϓϩʔνʢΊͬͪΌཁ໿ʣ • ϝδϟʔϦʔάͷ੒੷σʔλΛ࢖ͬͯ 1.࠷ۙ๣୳ࡧܥͷΞϧΰϦζϜͰ͍ۙબख୳͠ 2.֬཰෼෍ʢͬΆ͍ʣํ๏Ͱ༧ଌ੒੷Λ࡞Δ • ↑ͷ݁ՌΛStreamlitͰՄࢹԽ

Slide 12

Slide 12 text

ͳͥϝδϟʔͷσʔλͳͷ͔ • 3,000ຊ҆ଧୡ੒ऀ, ೔ຊϓϩ໺ٿ͸Ұਓ͔͍͠ͳ͍ʢ࡯͠ʣ   ※ʮ୭Ͱ͔͢ʁʯ࣭ͬͯ໰͸׃΍ͧ • ϝδϟʔ͸େਖ਼ٛΠνϩʔ༷ଞ, 3,000ຊ҆ଧୡ੒ऀ͕ଟ͍. • σʔλͷϥΠηϯε໰୊&εΫϨΠϐϯάͱ͔͸େมͰ͠ΐ.

Slide 13

Slide 13 text

ࡔຊ༐ਓʹ͍ۙϝδϟʔϦʔΨʔ ࢲʢshinyorkeʣ࡞੒, ʮzobristʯϞσϧͰग़ͨ݁͠Ռʢ΄΅ANNͰ͢ʣ ϝδϟʔϦʔάΛ஌ͬͯΔਓ͔ΒΈΔͱೲಘͷ݁ՌͩͱࢥΘΕ ໊͓લνʔϜ ੒੷ʢ௨ࢉʣ ଧ཰ຊྥଧ௨ࢉ҆ଧ ಛ௃ͱ͔ 9BOEFS#PHBFSUT   ʢ3FE4PYʣ ଧ੮͸ӈଧ ௨ࢉ014 ݱ໾۶ࢦͷ௒߈ܸܕγϣʔτ %FSFL+FUFS ʢ:BOLFFTʣ ଧ੮͸ӈଧ આ໌ෆཁͷελʔ ಺໺खʹݶΔͱ௨ࢉ҆ଧҐ 5SPZ5VMPXJU[LJ ʢ3PDLJFT FUDʜʣ ଧ੮͸ӈଧ ௨ࢉ014 ௒߈ܸܕγϣʔτ ͳ͓ຊڌ஍ +JNNZ3PMMJOT ʢ1IJMMJFT FUDʜʣ ଧ੮͸྆ଧ कඋܕͳγϣʔτ ࣮ಇ೥ͷແࣄ೭໊അ

Slide 14

Slide 14 text

σϞ͠·͢

Slide 15

Slide 15 text

ࡔຊ༐ਓͷࠓޙ - ҆ଧɾຊྥଧɾଧ఺ ࣅ͍ͯΔϝδϟʔϦʔΨʔXਓͷ੒੷Λ75%λΠϧͰࢉग़

Slide 16

Slide 16 text

ࡔຊ༐ਓͷࠓޙ - ଧ཰ ࣅ͍ͯΔϝδϟʔϦʔΨʔXਓͷ੒੷Λ75%λΠϧͰࢉग़

Slide 17

Slide 17 text

ࡔຊ༐ਓͷࠓޙΛ·ͱΊΔͱ 2027೥ʢ38ࡀʣ·Ͱنఆଧ਺֬อͰ͖ΔͬΆ͍.   ※نఆଧ਺͸443ଧ੮ʢ2019೥ͷࢼ߹਺143×3.1Ͱܭࢉ, ࢛ࣺޒೖʣ ೥౓ ೥ྸ ଧ਺ ҆ଧ ຊྥଧ ଧ఺ ଧ཰

Slide 18

Slide 18 text

ࡔຊ༐ਓબख, ௨ࢉ੒੷ʢ༧ଌ஋ʣ ͜ΕͰγϣʔτͩͬͨΒڧ͗͢Ͱ͸ʢ਒͑ʣ ظؒ ଧ਺ ҆ଧ ຊྥଧ ଧ఺ ଧ཰ ೥·Ͱ ˞ݱ࣮ ˞༧ଌ੒੷ ௨ࢉʢ༧ଌʣ

Slide 19

Slide 19 text

ߟ࡯ • 39ࡀ͝Ζʹ3,000ຊ҆ଧୡ੒…ͷϖʔε͸·͋·͋͋Γͦ͏.   ͨͩ͠ྼԽ͸ආ͚ΒΕͳ͍. • 36ࡀ͔ΒͷٸܹྼԽ͸कඋҐஔมߋͱ͔ͰઌԆ͹͠Ͱ͖ͦ͏.   ʲࢀߟʳѨ෦৻೭ॿ͸36ࡀ͔Βัख->ϑΝʔετʹίϯόʔτ • ௨ࢉຊྥଧʢ༧ଌʣ321ຊ…334ຊ͸ߦͬͯཉ͍͚͠ͲͲ͏͔

Slide 20

Slide 20 text

͜ͷ࿩͸͓͠·͍Ͱ͢…͕ʂʁ • ༧ଌϞσϧ࡞Γ͸ηΠόʔϝτϦΫεແ͠Ͱ͸Ͱ͖ͳ͔ͬͨ • ʮRʹΑΔηΠόʔϝτϦΫεೖ໳ʯ΋ग़ͨ͠,   ͜ͷลΛಛ௃ྔΤϯδχΞϦϯάతʹৼΓฦΓ͍ͨ • ͍ͬͯ͏ϩϯάτʔΫ͕Ͱ͖ͨΒ͍͍ͳ⚾   ʢҙ༁ɿࠓճ୺ંͬͨϞσϧͷ࿩Λ͍ͨ͠ʣ ӡӦͷօ༷, ͝ݕ౼ΑΖ͓͘͠ئ͍͠·͢

Slide 21

Slide 21 text

ήʔϜηοτ⚾ ͝ਗ਼ௌ͋Γ͕ͱ͏͍͟͝·ͨ͠. Shinichi Nakagawa(Twitter/Facebook/etc… @shinyorke)