Upgrade to Pro
— share decks privately, control downloads, hide ads and more …
Speaker Deck
Features
Speaker Deck
PRO
Sign in
Sign up for free
Search
Search
プロテニスにおいて疲れが勝敗に与える影響を定量化してみる
Search
MIZUTANI RYOTA
November 02, 2019
2
1.2k
プロテニスにおいて疲れが勝敗に与える影響を定量化してみる
Sports Analyst Meetup #5(
https://spoana.connpass.com/event/148275/)で発表したLT資料です
。
MIZUTANI RYOTA
November 02, 2019
Tweet
Share
More Decks by MIZUTANI RYOTA
See All by MIZUTANI RYOTA
言語モデルにおける推論パラメータと小説生成への適用について
rmizuta3
0
610
PythonユーザによるRust入門
rmizuta3
16
7.7k
Featured
See All Featured
How to Think Like a Performance Engineer
csswizardry
25
1.8k
Performance Is Good for Brains [We Love Speed 2024]
tammyeverts
10
1k
Principles of Awesome APIs and How to Build Them.
keavy
126
17k
BBQ
matthewcrist
89
9.8k
Automating Front-end Workflow
addyosmani
1370
200k
Fantastic passwords and where to find them - at NoRuKo
philnash
51
3.4k
Speed Design
sergeychernyshev
32
1.1k
A designer walks into a library…
pauljervisheath
207
24k
実際に使うSQLの書き方 徹底解説 / pgcon21j-tutorial
soudai
PRO
183
54k
Keith and Marios Guide to Fast Websites
keithpitt
411
22k
GraphQLとの向き合い方2022年版
quramy
49
14k
10 Git Anti Patterns You Should be Aware of
lemiorhan
PRO
656
60k
Transcript
ϓϩςχεʹ͓͍ͯ ർΕ͕উഊʹ༩͑ΔӨڹΛ ఆྔԽͯ͠ΈΔ 4QPSUT"OBMZTU.FFUVQ ਫ୩྄ଠ !SNJ[VUB
ࣗݾհ • ࢯ໊ɿਫ୩྄ଠ(@rmizuta3) • ͓ࣄɿϚʔέςΟϯάܥاۀͰσʔλ׆༻Λߟ͑Δਓ • झຯɿσʔλੳɺςχε • εϙʔπྺɿ •
ςχεྺ15 • େֶ࣌ମҭձॴଐ • ;͡ΈࢢࢢຽେձγϯάϧεBڃ̏Ґ
ʮۋ৫ർΕ͍ͯͨʯ • ݄̒ͷશถΦʔϓϯ४ʑܾউͰ6-1, 6-1, 6-3Ͱۋ৫Λഁͬ ͨ࣌ͷφμϧͷίϝϯτ • ४ʑܾউ·Ͱͷ4ࢼ߹ͷ߹ܭࢼ߹࣌ؒφμϧͷ9࣌ؒ ʹର͠ɺۋ৫13࣌ؒΛ͍͑ͯͨɻ ग़యɿIUUQTOFXTUFOOJTOFUOFXTUPEBZIUNM
ςχεͷࢼ߹ͷಛ • ࢼ߹͕࣌ؒৼΕ෯͕େ͖͍ • 4େେձͰ̑ηοτϚον̏ηοτઌऔ • 1࣌ؒڧͰऴΘΔ͜ͱ͋Ε5࣌ؒҎ্͔͔Δ߹͋Δ ͷ̐େେձͷࢼ߹࣌ؒ
ςχεͷࢼ߹ͷಛ • େձͷࢼ߹ִ͕ؒ͘ɺଟ͍ɻ • େձ։࠵ظؒ̎िؒɻͦͷதͰ࠷େ̓ࢼ߹Λઓ͍ ൈ͘ඞཁ͕͋Δ ճઓ
ճઓ ճઓ ճઓ ճઓ ճઓ ճઓ ճઓ શͷఔ ճઓ ४ʑܾউ ४ʑܾউ ४ܾউ % ४ܾউ 4 ܾউ % ܾউ 4
ർΕͷӨڹ͋Γͦ͏ɻ ͕࣮ͩࡍͲͷ͘Β͍ͷӨڹ͕ ͋ΔͷͩΖ͏͔ʁ
ੳํ • ϩδεςΟοΫճؼΛ༻͍ͯർΕ͕উʹ༩͑ΔӨڹΛݕূ • ճؼࣜ y = 1 1 +
exp( − (a1 − a2 )x1 + (b1 − b2 )x2 ) ɹɿউഊ PS ɹɿબखͷڧ͞ɿબखͷڧ͞ ɹɿબखͷർ࿑ɹɹɿબख̎ͷർ࿑ a1 a2 b2 b1 y
બखͷڧ͞ • ΠϩϨʔςΟϯάͷσʔλΛ༻ • https://ultimatetennisstatistics.com/ ͷΠϩϨʔςΟϯά
બखͷർ࿑ • ֤ࢼ߹ͷࢼ߹࣌ؒͷσʔλΛར༻ • https://github.com/JeffSackmann/tennis_atp • େձظؒதɺࢼ߹ͷർΕੵ͢Δ͕ɺճ෮ͷྔߟྀ ͍ͨ͠
બखͷർ࿑ͷߟ͑ํ ࢼ߹࣌ؒ ർ࿑ ճઓ ճઓ
ճઓ ർ࿑લࢼ߹ͷࢼ߹࣌ؒ ɹɹɹɹલࢼ߹ͷർ࿑ ർ࿑ଘ ଘ͕ͷ߹
ർ࿑ଘͷࢉग़ • ർ࿑ଘΛมԽͤ͞ͳ͕ΒϩδεςΟοΫճؼΛ࣮ࢪ • ༧ଌͱ࣮ࡍͷͷRMSE͕࠷খʹͳΔͷΛબ ർ࿑ଘ͕̔ͷ࣌ ࠷ͯ·Γ͕ྑ͍
ϩδεςΟοΫճؼͷ࣮ߦ • 2018ͷ4େେձͷ482ࢼ߹ͷσʔλΛ༻ y = 1 1 + exp( −
(a1 − a2 )x1 + (b1 − b2 )x2 ) ɹɿউഊ PS ɹɿબखͷڧ͞ɿબखͷڧ͞ ɹɿબखͷർ࿑ɹɹɿબख̎ͷർ࿑ a1 a2 b2 b1 y
݁Ռ ർ࿑ʹ͕ࠩ͋Δͱɺ ͕ࠩͳ͍߹ͱൺֱͯ͠উ͕ഒʹ ճؼ Φοζൺ
݁Ռͷαϯϓϧ ϕʔεͷۋ৫WTφμϧͷউͱർ࿑ͷࠩͷؔ
·ͱΊ • ςχεʹ͓͍ͯർΕ͕উʹ༩͑ΔӨڹΛϩδεςΟΫճ ؼΛ༻͍ͯఆྔԽͨ͠ • ർ࿑ͷ͕ࠩ100͋Δ߹ɺ͕ࠩͳ͍߹ͱൺֱͯ͠উ ͕60%ʹݮগ͢Δ͜ͱ͕Θ͔ͬͨ • ଥੑͷ֬ೝͷͨΊผͷΞϓϩʔνͰͬͯΈ͍ͨ