Upgrade to Pro
— share decks privately, control downloads, hide ads and more …
Speaker Deck
Sign up for free
Menu
Search
Features
All features
Private URLs
Password Protection
Custom URLS
Scheduled publishing
Remove Branding
Restrict embedding
Deck Collections
Notes
Features
All features
Private URLs
Password Protection
Custom URLS
Scheduled publishing
Remove Branding
Restrict embedding
Deck Collections
Notes
Explore
Featured decks
Featured speakers
Programming
Technology
Storyboards
Explore
Featured decks
Featured speakers
Programming
Technology
Storyboards
Pricing
Search
Sign in
Sign up for free
BigQueryとPythonではじめるプロ野球選手の成績予測(もしくは成績占い) / Bas...
Search
Shinichi Nakagawa
PRO
May 27, 2022
Research
4.6k
0
Share
Embed
Copy iframe code
Copy JS code
Copy link
Start on current slide
BigQueryとPythonではじめるプロ野球選手の成績予測(もしくは成績占い) / Baseball Player Performance Prediction using BigQuery and Python
Baseball Play Study mini 2022/5/27 登壇資料
Shinichi Nakagawa
PRO
May 27, 2022
More Decks by Shinichi Nakagawa
See All by Shinichi Nakagawa
LLMの出力を"いい感じに"する技術 / Taming LLM Output: AI Agent Design Patterns with FastAPI
shinyorke
PRO
4
2.2k
野球解説AI Agentを開発してみた - 2026/02/27 LayerX社内LT会資料
shinyorke
PRO
0
740
WBCの解説は生成AIにやらせよう - 生成AIで野球解説者AI Agentを実現する / Baseball Commentator AI Agent for Gemini
shinyorke
PRO
1
750
自らを強いエンジニアにするための3つの習慣 2025/ Fitter happier more productive
shinyorke
PRO
0
330
生成AI時代におけるSREの進化とキャリア戦略 / Building an Embedded SRE team and my career
shinyorke
PRO
0
190
生成AIを活用した野球データ分析 - メジャーリーグ編 / Baseball Analytics for Gen AI
shinyorke
PRO
1
6.8k
ゼロから始めるSREの事業貢献 - 生成AI時代のSRE成長戦略と実践 / Starting SRE from Day One
shinyorke
PRO
3
9.8k
AI・LLM事業部のSREとタスクの自動運転
shinyorke
PRO
0
590
実践Dash - 手を抜きながら本気で作るデータApplicationの基本と応用 / Dash for Python and Baseball
shinyorke
PRO
2
5k
Other Decks in Research
See All in Research
適応的スパムフィルタのための軽量な類似メッセージカウンタ / jsai2026-adaptive-spam-filter
monochromegane
0
5.9k
SAM3を用いたコマ・吹き出しの 領域検出と分割構造からの読み順推定
kzmssk
0
180
論文紹介: Understanding Epistemic Language with a Language-augmented Bayesian Theory of Mind
hisaokatsumi
0
190
Développer des solutions de réduction des émissions de méthane entérique : 1ers résultats Méthane 2030
institutdelelevage
PRO
0
240
敵対生成プロンプト同時探索による内省型プロンプト最適化
kinoue_smarthr
0
440
医療LLMの現在地〜最新研究から社会実装までを考える〜
kento1109
1
1.8k
Cross-Media Human-Information Interaction
signer
PRO
0
270
Evaluation génomique des femelles laitières croisées : comprendre la méthode pour accompagner les éleveurs à son utilisation
institutdelelevage
PRO
0
160
[SNLP2026] Scaling up Test-Time Compute with Latent Reasoning: A Recurrent Depth Approach
wataruuuuu
0
330
20260624 NLP colloquium: 単一のhubテキストがCLIPを壊す:hubnessによる埋め込みの脆弱性特定
de9uch1
2
290
HackSick vol.7 LT資料【LLMアーキテクチャ入門・事前学習時の躓き所解説】 スパースなAttention・状態空間モデル
rikkabotan7
0
210
最先端NLP勉強会2026 論文紹介:Reasoning with Sampling: Your Base Model is Smarter Than You Think (ICLR 2026 paper)
kogoro
4
690
Featured
See All Featured
The Straight Up "How To Draw Better" Workshop
denniskardys
239
140k
Intergalactic Javascript Robots from Outer Space
tanoku
273
27k
Mozcon NYC 2025: Stop Losing SEO Traffic
samtorres
1
570
Technical Leadership for Architectural Decision Making
baasie
3
600
How to Align SEO within the Product Triangle To Get Buy-In & Support - #RIMC
aleyda
2
1.8k
Tips & Tricks on How to Get Your First Job In Tech
honzajavorek
1
790
Rebuilding a faster, lazier Slack
samanthasiow
85
9.7k
The Invisible Side of Design
smashingmag
301
52k
Information Architects: The Missing Link in Design Systems
soysaucechin
1
1.2k
Are puppies a ranking factor?
jonoalderson
2
4k
Discover your Explorer Soul
emna__ayadi
2
1.3k
<Decoding/> the Language of Devs - We Love SEO 2024
nikkihalliwell
1
350
Transcript
ಥવͰ͕͢””͍͍ͬͯͰ͔͢? Shinichi Nakagawa@shinyorke Baseball Play Study mini 2022/05/27
ຊͷςʔϚʮٿͱAIͱٕज़ʯ • ٿAIΛ࡞ΔͨΊͷηΠόʔϝτϦΫεͱΞϧΰϦζϜ • ٿAIΛࢧ͑Δٕज़ - PythonͱGoogle CloudΛఴ͑ͯ • ٿAIͰ͍·͢,
ʮਪ͠ͷબखͷ5ઌʯ ͖͏ͷਓೳΛ࡞ͬͨͷͰҰॹʹ༡΅͏ͥ⽁ʢཁʣ
ࠓճͷ͍ʮଧऀͷʯͰ͢ खVer.ແ͘ͳ͍Ͱ͕͢, ݁Ռ͕ඍົͩͬͨͷͰଧऀͷΈͰΒ͍͖ͤͯͨͩ·͢🙏
Who am I ?ʢ͓લ୭Αʣ • Shinichi Nakagawaʢத ৳Ұʣ • େͷSNSͰʮshinyorkeʢ͠ΜΑʔ͘ʣʯͱ໊͍ͬͯ·͢
• ΞΫηϯνϡΞגࣜձࣾϚωʔδϟʔʢຊۀʣ • ຊۀͷํͰʮGoogle Cloudڧ͍ϚϯʯతͳཱͪҐஔͰ ιϦϡʔγϣϯΞʔΩςΫτʢSRE/DevOpsपΓʣ • ݸਓ׆ಈʮੜͷٿσʔλαΠΤϯςΟετʯͱͯ͠ ٿʹؔ͢ΔσʔλαΠΤϯεͱΤϯδχΞϦϯάΛ͍ͯ͠·͢ ʢ͔ͭ, ຊۀͰػցֶशΤϯδχΞɾσʔλαΠΤϯςΟετܦݧ͋Γʣ • ւಓຊϋϜϑΝΠλʔζ&ΦʔΫϥϯυɾΞεϨνοΫεͷϑΝϯ⽁
ຊͷଧॱ • ٿAIΛ࡞ΔͨΊͷηΠόʔϝτϦΫεೖ • ٿAIΛ࣮͢Δ - Google CloudΛఴ͑ͯ • ಥવͰ͕͢””͍͍ͬͯͰ͔͢⽁
- ٿAIͰ
ٿAIΛ࡞ΔͨΊͷηΠόʔϝτϦΫεೖ
ٿAIΛࢧ͑ΔηΠόʔϝτϦΫε • ηΠόʔϝτϦΫε #ͱ • ηΠόʔϝτϦΫεͷ͖΄Μ • ʮͦͬ͘Γ͞Μʯ͔Β༧ଌ - PECOTAϞσϧ
• shinyorke’s༧ଌϞσϧʮzobristʯվΊʮkenshiʯ શ෦͢ͱ3.34͔͔࣌ؒΔͷͰࠓ֓ཁͷΈհ🐯
ηΠόʔϝτϦΫε #ͱ • ٿʹ͓͍ͯൃੜ͢ΔσʔλΛ౷ܭֶతͳΞϓϩʔνͰੳΛߦ͍, ʮબखͷೳྗʯʮνʔϜͷڧ͞ʯͳͲநతͳ֓೦ΛఆྔతʹࢦඪԽ͠, νʔϜɾબखɾϑΝϯʹཱͯΔͨΊͷՊֶతͳΞϓϩʔνɾߟ͑ํͷ͜ͱ. • Ҏલσʔλ͕ओྲྀ͕ͩͬͨ,
ʮελοτΩϟετʯʮτϥοΫϚϯʯͱ͍ͬ ͨ, ܭଌػثτϥοΩϯάσʔλΛ༻͍ͯߦ͏ͷ͕ࠓͷτϨϯυ • ͳ͓, σʔλͷΈͰे͗͢Δ͙Β͍໘ന͍ࣄ͕ग़དྷ·͢ ʢ㲈τϥοΩϯάσʔλٿɾٕज़ڞʹઐ͕ࣝΘΕΔ&қߴ͍ʣ
ηΠόʔϝτϦΫεͷ͖΄Μ • ηΠόʔϝτϦΫεγϯϓϧͳ࢛ଇԋࢉ͓Αͼ౷ܭͰߦ͏ࣄ͕Ͱ͖Δ • Α͘ΒΕ͍ͯΔʮOPSʯʮWHIPSʯͳͲ, ࣜͦͷͷిExcelͰܭࢉ͕Մೳʢ㲈ϓϩάϥϛϯάෆཁʣ • Ұํ, ʮWARʯʮRCʯͳͲͷࢦඪܭࢉׂ͕ͱෳࡶ,
ExcelͰग़དྷͳ͘ແ͍͕, ϓϩάϥϛϯάSQL, ػցֶशͳͲͰॲཧͨ͠΄͏͕ྑ͍ύλʔϯଘࡏ͢Δ. • ༧ଌʮಛྔΤϯδχΞϦϯάʯͱͯ͠ѻ͏ͱ࣮ݱ͢Δ͜ͱ͕Ͱ͖Δʂ ʢͱ͍͏ͷ͕͜ͷൃදͷٕज़తͳςʔϚͰ͢ʣ
ٿσʔλΛಛྔʹม͢Δ ಛྔʢ㲈ʣʹมɾ୯ҐΛἧ͑Δಓͳ࡞ۀ͕ඞཁ • -> • ͦͷ··͑ΔϞϊ͕ଟ͍. ྫ͑҆ଧ, ࢛ٿ,
ࡾৼͳͲ. • Θ͔Γ͍͢୯Ґʹਖ਼نԽɾεέʔϦϯά͢Δ. RC, wRAA, wOBAͳͲͷηΠόʔϝτϦΫεࢦඪ. • Ҏ֎ͷσʔλ -> • ར͖ଧ੮ͷࠨӈ, ఱવࣳ or ਓࣳ, ֎ or υʔϜ or େࣗવʢॴͱݴ͍ͬͯͳ͍ʣ? • Ͱແ͍σʔλΛಛྔʹ͢ΔͨΊͷॲཧ͕ඞཁʢΧςΰϦʔʹ͢ΔͳͲʣ
ϓϩٿબखͷΛ͏ٕज़ • ༧ଌͦͷͷηΠόʔϝτϦΫεᴈ໌ظ͔Β͋Δఆ൪ωλͰ, ΞϝϦΧʹϑΝϯ͚ͷ༧ଌαΠτ͕͋Δ͙Β͍ͷΓ্͕Γ. • τϥοΩϯάσʔλ͕ओྲྀͷࠓͰબखͷ݈߁ཧύϑΥʔϚϯε ଌఆͳͲΛ௨ͯ͡কདྷͷύϑΥʔϚϯεΛ༧ଌ͢Δ͜ͱ. ※༧ଌͷݱ׆༻,
ͱݴ͑ΔʢPlayer’s Developmentʣ • ݹయత͔ͭදతͳ༧ଌϞσϧͱͯ͠, PECOTAʢϖίλʣ͕༗໊.
PECOTA - ࠷౷ܭతͳ༧ଌϞσϧ • 2003ʢ19લʣʹϦϦʔεͨ͠MLBͷ༧ଌϞσϧ • ʮաڈͷࣅ͍ͯΔબखͷʯ͔Β༧ଌΛࢉग़ ͳ͓۩ମతͳख๏ɾࣜඇެ։ʢߟ͑ํͪΒ΄Βॻ͍ͯ͋Δʣ •
ޙʹ2008ถࠃେ౷ྖબڍͷউऀΛ49/50भతதͤͨ͞ ౷ܭֶऀωΠτɾγϧόʔ͕։ൃ ※ؾʹͳΔํʮγάφϧ&ϊΠζʯͱ͍͏ॻ੶ΛಡΜͰ͍ͩ͘͞
ʮͦͬ͘Γ͞ΜʯΛ୳͢ࣄͰΛ༧ଌͰ͖Δ!? աڈϝδϟʔϦʔάͰσϏϡʔͨ͠બख2021·Ͱʹ20, 370ਓ͍ΔʢLahman’s Baseball Databaseௐʣ ͜Ε͚͍ͩΕ, ʮੲͷ͋ͷਓͬΆ͍ʯ͙Β͍ग़ͤΔͷͰͳ͔Ζ͏͔???
shinyorke’s༧ଌϞσϧʮkenshiʯര • աڈʢؚΉݱʣʹଘࡏͨ͠ٿબखͷΛݩʹ, ʮࣅ͍ͯΔબखΛΫϥελϦϯάʯ Ͱ͖ͨΒ༧ଌ࡞ΕΔͷͰ? -> ࣮PECOTA͜ͷΞϓϩʔνʢ࠷ॳظʣ • ηΠόʔϝτϦΫεͰʮྨࣅੑείΞʯͱ͍͏ࣅ͍ͯΔબखΛಋ͕ࣜ͋͘Δ͕,
೦ͳ͕Β͕ܽଟ͍ʢ௨ࢉͰΫϥελϦϯά͢ΔͨΊ, όΠΞε͕ڧ͘ग़Δʣ • ʮ͋ΔಛྔΛݩʹΫϥελϦϯάʯಛྔΤϯδχΞϦϯά͕ಘҙͳλεΫ ػցֶशతͳΞϓϩʔνͰߦ͚ΔͷͰ?આ -> Ͱ͖ͨ🙌 • ϝδϟʔϦʔά൛AIʮzobristʯΛ։ൃ -> ͍͍ײͩͬͨ͡ͷͰຊϓϩٿ൛Λ࡞ˡࠓ͜͜ ͜͏ͯ͠, shinyorke’sϓϩٿ༧ଌϞσϧ&ٿAIʮkenshiʯ͕ര.
ٿAIΛ࣮͢Δ - Google CloudΛఴ͑ͯ
ٿAIʮkenshiʯΛ࣮͢Δ • ΞʔΩςΫνϟͷશମ૾ • σʔλऔಘͱલॲཧ • ΞϧΰϦζϜΛܾΊͯΫϥελϦϯά • ༧ଌΛ፻͢Δੜ͢Δ ͪͳΈʹkenshi໊ͬͯલͷ༝དྷӈ྆ଧͷ͋ͷબखΑΓʢࠢʣ
ϓϩٿબख༧ଌϓϩμΫτશମ૾
ΞʔΩͷجຊํ • σʔλͯ͢BigQueryʹूΊΔʢ㲈Google CloudͰͯ͢ΛݻΊͨཧ༝ʣ • ֶशσʔλςετσʔλͯ͢BQ • ޙड़͢ΔલॲཧɾσʔλཧΛͳΔ͘SQLͰΓ͔ͨͬͨ • αʔόϨεͳαʔϏεΛத৺ʹબΜͰ͏ʢ㲈VMͰ͋Δඞཁ͕ແ͍ʣ
• ΞϓϦΫϩʔϥʔCI/CDαʔόϨεܥͷαʔϏεͰݻΊΔ • ʮͬͨʯ͚ͩඅ༻ʹͳΔͷͰࡒʹ༏͍͠&εέʔϦϯάָ
αϥοͱղઆ • Data Analytics • BigQuery͕ͯ͢ͷத৺, σʔλͯ͢͜͜ • Cloud ConsoleͰΫΤϦʔΛॻ͍ͯσʔληοτ࡞,
͍͠λεΫJupyter Lab্Ͱ࣮ࢪ • ωοτ͔Βऩू͢ΔσʔλʢCSVʣCloud Storageʹอଘ, Cloud FunctionsΛͬͯBigQueryʹExport • Web App • StreamlitʢޙͰղઆʣͰ࣮ͨ͠ΞϓϦΛCloud RunͰϗετ • CI/CDGitHub ActionsͰαΫοͱ
ϗϯτʹࡉ͔͍ٕज़ղઆϒϩάͰ https://shinyorke.hatenablog.com/entry/cloud-arch-serverless ࠓճͷൃද༻ͷ͓ֆ͔͖Ͱ͕ͨ͠ϒϩάͰόζͬͨ&ผͰৄͤ͘͠Εʂ
σʔλͷऔಘ • ֶशσʔλϝδϟʔϦʔάͷσʔλΛ༻ • Lahman’s Baseball Database • ্هσʔλϕʔεͷCSVσʔλΛBigQueryʹimport •
ϓϩٿͷσʔλBaseball Reference͔ΒεΫϨΠϐϯά • 2021γʔζϯऴྃ࣌ΛݩʹεΫϨΠϐϯά • Pythonͷrequests-htmlͰΫϩʔϥʔΛ࣮, CSVอଘ -> BigQuery
ϝδϟʔϦʔάͷσʔλΛͬͨཧ༝ • ຊͷϓϩٿͰ·ͱ·ͬͨσʔληοτ͕ଘࡏ͠ͳ͍ • ͋Δॴʹ͋Δ͕, ݖརతʹ͑Δ͔ո͍͠ • ϝδϟʔϦʔάΦʔϓϯσʔλ͕ॆ࣮͔ͭݖརେৎ • αϯϓϧσʔλͷେ͖͞ʢ100Ҏ্͋ΔͷͰे͗͢Δʣ
• ಉ͡ٿͱ͍͏ڝٕ͔ͭهมΘΒͳ͍ͷͰӨڹগͳ͍ͱஅ
લॲཧ • ֶशʹඞཁͳσʔλSQLͰՃ, Viewʹͯ͠อଘ →Google Cloud ConsoleͰ࣮ࢪ • Ͳ͏ͯ͠SQLͰ໘͍͘͞ͷΛPandasͳͲͰॲཧ
→ࣗͷPC্ʹ࡞ͬͨJupyter LabڥͰ࣮ࢪ • ֶशʹඞཁͳ௨ࢉɾผΛࢉग़͢ΔͨΊͷ ΫΤϦʔσʔληοτΛͻͨ͢Β࡞Γ·ͬͨ͘
લॲཧͷྫ - SQLͰߦ͏ٿͷಛྔநग़ • ଧ, ग़ྥ, OPSతͳͷ SQLͰܭࢉͰ͖Δ. •
͏ͪΐͬͱෳࡶͳࢦඪ. ྫ͑wOBAͱ͔. • ্هBigQueryͰ݁͠·ͨ͠.
લॲཧͷྫ - SQLͰߦ͑ͳ͍ͷ? • ࡶͳॲཧɾܭࢉ͕ೖͬͨΓ, ߦྻͰ·ͱ·ͬͨϞϊͷॲཧ PythonRͰॲཧ͕ϕετ. •
ྫ͑ϐϘοτςʔϒϧ, άϧʔϐϯάͳͲPandasͰ ॻ͍ͨ΄͏͕Θ͔Γ͍͢ ͱ͖͋Δʢॾઆ͋Γ·͢ʣ • SQL͕ۤखͳํશ෦ͬͪ͜Ͱͬͯྑ͍͔.
ΞϧΰϦζϜΛܾΊͯΫϥελϦϯά • ʮࣅ͍ͯΔબखʯΛྨ͢ΔλεΫ • ΞϧΰϦζϜΛܾΊΔˠ࠷ऴతʹANNʹ • AnnoyʢΞϊΠʔʣͰരANN ྨλεΫΛ࡞Γ, ςετΛॻ͖, γϡοͱCIͰ࠶࣮ߦՄೳʹ.
ࣅ͍ͯΔબखΛ୳͢ɾྨ͢Δ • ௨ࢉͱकඋҐஔ͝ͱͷग़ճΛಛྔͱ͢Δ͜ͱʹΑΓ, ʮࣅ͍ͯΔબखʯΛ୳͢͜ͱ͕ՄೳͳͷͰ? • ಛྔΛͬͯΫϥελϦϯάͯ͠ڑΛܭଌ, ͍ۙॱͰϥϯΩϯάԽ͢ΔʢϢʔΫϦουڑͳͲͰʣ •
ͳ͓, PECOTAʢ͓ͦΒ͘ʣߟ͑ํಉ͡.
ANNʢۙࣅ࠷ۙ୳ࡧʣΛ࠾༻ • ग़ࢼ߹, ଧ੮, ओཁͳଧܸʢ҆ଧ, ຊྥଧ, ଧ, etc…ʣ • कඋҐஔʢશ9ϙδγϣϯ,
DHߟྀ͠ͳ͍ʣผͷग़ճ • ্هΛಛྔͱͯ͠ANNʢۙࣅ࠷ۙ୳ࡧʣΛ͔ͭͬͯ ϢʔΫϦουڑΛࢉग़͠, ͍ۙબखΛूΊΔ͜ͱʹ. • ʮAIʹΑΔࣆδϟύϯબग़ʯͱ͍͏ωλͰར༻->݁Ռ্ʑ https://shinyorke.hatenablog.com/entry/tokyo2020-samurai-japan • ࣮Annoyͱ͍͏ศརͳϥΠϒϥϦΛ͍·ͨ͠.
AnnoyΛͬͨANNʹΑΔΫϥελϦϯά. ूΊͨσʔλΛ͠ࠐΜͰΔ͜ͱͰΫϥελϦϯά͕Ͱ͖·ͨ͠.
݁ՌΛݟͯΈͨ • ΦϦοΫε٢ాਖ਼ঘʹࣅ͍ͯΔਓΛ୳͢ • ༧ଌϞσϧʹ٢ాਖ਼ঘͷΛͯ͠ ΫϥελϦϯά݁ՌΛௐࠪ • ϋϯΫɾΞʔϩϯ, ΟϦʔɾϝΠζ,
ήϨʔϩଞ, ࣅ͍ͯΔ֎ख͕औΕͨͷͰ ޭͱݴ͑ͦ͏🎉 ※ήϨʔϩڈΦΦλχαϯͱHRԦ૪͍ͨ͠ήϨʔϩJr.ͷ͓͞Μ
༧ଌͷग़͠ํ • ΫϥελϦϯάͷ݁Ռ, ্ҐʹϥϯΩϯά͞Εͨબखͷ ྸผΛऔಘ • ྸผͷฏۉύʔηϯλΠϧΛࢼͯ͠, ऩ·Γͷྑ͍ࣈʹ͢Δ
• ଧɾ҆ଧͳͲʮੵΈॏͶʯͷΛ༧ଌޙ, ଧͳͲͷʮʯΛද͢Λܭࢉ
ϓϩμΫτʹ͢Δ • ͻͱ·࣮ͣݧతͳΞϓϦέʔγϣϯΛ StreamlitͰ࣮ • StreamlitҰݴͰݴ͏ͱ ʮJupyter notebookΛΞϓϦʹ͢Δʯ
ͨΊͷFramework • Dockerίϯςφʹͯ͠ Cloud RunͰϗεςΟϯά
ಥવͰ͕͢””͍͍ͬͯͰ͔͢?
AIͰ͏ʮࠓ, ؾʹͳΔϓϩٿબखʯ • ݱࡏઈௐ, ޥͷ͋ͷਓ • ೋ಄ཽʢೋྲྀʣͤ͞Δඞཁ͋Δͷ͔ແ͍ͷ͔? • BIG BOSSʹࣅͯΔʢ͔͠Εͳ͍ʣ͋ͷબख
ຊ12ٿஂ৮Ε͍ͨ…Ͱ͕࣌ؒ͢ͷ߹ʹΑΓ🙏
ઈௐͳޥͷ͋ͷਓͱ͍͑ • ࡔ কޗʢౡʣ - 2016υϥϑτ4Ґ • ࡢ͍ͭʹϒϨΠΫ, ࠓ͜͜·Ͱଧരൃ •
ϝΠϯัख͕ͩकΕΔϢʔςΟϦςΟ
͜ΕΤά͍ະདྷ༧ਤʢੌʣ
ࡔ কޗબखͷະདྷ • ࠓͷ༧ʮଧ.309 ຊྥଧ20ຊ ଧ70 OPS .903ʯ • ڈͷงғؾ͔Β͢ΔͱϦΞϧʹୡՄೳͳ༧ײ͕!?
• ݸਓతʹͬͺัखͬͯ΄͍͠, νʔϜࣄͳΜ͚ͩΕͲ ʢଧͯΔัख͍Δ͚ͩͰΞυόϯςʔδେ͖͍ʣ
ཽͷະདྷΛ͏ - ೋਓͷཽઓ࢜ • AɾϚϧςΟωεʢதʣ - 2018ೖஂ • ࠜඌ ߉ʢதʣ
- 2018υϥϑτ1Ґ • ଧ͓ΑͼೋྲྀͰͪΐͬͱΛݺΜͰ͍ΔೋਓΛ͏
ཽͷະདྷ໌Δ͍͔?
AɾϚϧςΟωεબखͷະདྷ • ࠓͷ༧ʮଧ.290 ຊྥଧ10ຊ ଧ44 OPS .862ʯ • OPSҎ֎ຊؾͰୟ͖ग़ͦ͠͏ͳࣈͳؾ͕͢Δ? •
ཉΛݴ͑֎͡Όͳͯ͘ัखͰग़ͯ΄͍͠ ัखͰ͜Ε͚ͩଧͬͨΒࠓͷٿͩͱੌ͍͜ͱʹ
ͳΔ΄Ͳ?
ࠜඌ ߉બखͷະདྷ • ࠓͷ༧ʮଧ.244 ຊྥଧ1ຊ ଧ6 OPS .654ʯ 5ޙ·Ͱͷ༧ଌ…͏ʔʔΜ?
• ඇৗʹौ͍ධՁ, ೋྲྀΛࢼ͢ҙຯ༧ଌ͚ͩͩͱ͋Γͦ͏? • ൩ܕͱ৴͍ͨ͡, ͍͘ΒͳΜͰٿAIͷධՁ͕ौ͗͢? ͪͳΈʹ౻ݪ ګେʢϩοςʣͳ͔ͳ͔ौ͍ධՁʹ
BIG BOSSͷޙܧऀ୭ͩ? • ສ தਖ਼ʢຊϋϜʣ - 2018υϥϑτ4Ґ • ύϫʔͱεϐʔυ, ࡶ͞Λ݉Ͷἧ͑ͨϑΟδΧϧϞϯελʔ
ϑϧεΠϯάͰ͔ͬඈ͢ଧܸͱڧݞΛੜ͔ͨ͠कඋ ݱ࣌ͷBIG BOSSͦͷͷ • ࢲ, shinyorke͕ࠓ࠷ਪ͍ͯ͠Δϓϩٿબख
ࢥͬͨΑΓBIG BOSSͬΆ͞?
ສ தਖ਼બखͷະདྷ • ࠓͷ༧ʮଧ.252 ຊྥଧ18ຊ ଧ52 OPS .780ʯ • ϗϯτʹୟ͖ग़ͦ͠͏ͳࣈ,
ग़ػձ&ଧ࣍ୈͰ ͳΜͩͬͨΒຊྥଧ༧ଌ௨Γ͔ͨ͠͠Β͔͢? • 5ޙʹOPS.900͑Β͍͠ͷͰ, ͜ͷ͍ͨͬͯཉ͍͠
ͪͳΈʹ, ϓϩτλΠϓͰ࡞ͬͨ ผͷAIϞσϧ͕มͳ༧ଌͯ͠·ͨ͠ ʢࠓճVer.Ͱ͍͟͝·ͤΜʣ
ສ தਖ਼ͱBIG BOSS ଧ ຊྥଧ ଧ #*(#044ࡀ ʢɾࡕਆʣ
ຊ ଧ ສதਖ਼ࡀ ʢͷ༧ଌʣ ຊ ଧ ΊͬͪΌ৽ঙ߶ࢤબखΜʂʂʂ
͖͏AIͰBIG BOSSͷޙܧऀ, ݟ͚ͭ·ͨ͠ʢ͜ͳΈʣ
݁ͼ
͖͏AIͷ՝ͱ࣍ͷςʔϚ • ৽ਓબखͷ༧ଌ͕ʢϞσϧͷ্༷ʣͰ͖ͳ͍ • ݱϞσϧ௨ࢉϕʔεͰͷֶश&༧ଌͰ͋ΔͨΊ, ࣮ແ͍ϧʔΩʔͷ༧ଌ͕ग़དྷͳ͍ • ߴߍɾେֶͷΛͦͷ··͑…ͱ͍͏୯७ͳղܾ͕Ͱ͖ͳ͍ • Ҏ֎ͷઆ໌มΛՃ͍͑ͨ
• ͬͺΓτϥοΩϯάσʔλ͍͍ͨʂ͋ͱମ֨ͱ͔ • ຊϓϩٿͰΔखஈແ͍͕, ϝδϟʔϦʔάBaseball SavantͰ͍͚ΔͷͰҾ͖ଓ͖AIΛҭ͍͖͍ͯͯͨ • ख൛͕͋Μ·Γ͓͠Ζ͘ͳ͔ͬͨͷ͕չ͍͠ • ҰԠ͋ΔͷͰ͕͢, ඍົͩͬͨͷͰൃද߇͑·ͨ͠, ࠤʑ ࿕رͷະདྷ༧ਤݟͯΈ͍ͨͷͰ͏গ͕͠ΜΓ·͢. • ݱ࣮తʹ, ϓϩٿͰΓͳ͍ಛྔ͕͋Γ·ͯ͠…ϝδϟʔϦʔά൛े࣮༻ʹת͑ΔͷͰ͕͢😇
࣍ͷల։ • ͍ʢ༧ଌʣαΠτͷ্ཱͪ͛. σʔλͷݖརͱ͔ॾʑ্ख͘ղ্ܾͨ͠Ͱʢଟ͍͚Δͱࢥ͏ʣ. • ༧ଌΞϧΰϦζϜͷվྑ. ϝδϟʔϦʔά൛ͰτϥοΩϯάσʔλΛͬͨϞσϧͷ։ൃ. •
PyCon JP 2023ͱ͔, ϦΞϧ։࠵ͷBaseball Play StudyͰ·ͨձ͓͏.
ಥવͰ͕͢””͍͍ͬͯͰ͔͢? • ٿAIηΠόʔϝτϦΫεͱػցֶशͷԠ༻Ͱ࣮Մೳ • ٿAIBigQueryPythonͰ࡞ΕΔ • ัखัखΛΔ͖Ͱ, ϚϯνϡBIG BOSSͷޙܧऀ ·ͩ·ͩ༡΅͏ͱࢥ͍·͢ͷͰҾ͖ଓ͖ΑΖ͘͠ʂ
ήʔϜηοτ ͝ਗ਼ௌ͋Γ͕ͱ͏͍͟͝·ͨ͠ Shinichi Nakagawa(Twitter/Facebook/etc… @shinyorke)