投球を可視化する技術〜Analyzing Pitching Data With Python

Slide 1

Slide 1 text

No content

Slide 2

Slide 2 text

Who am I? • Shinichi Nakagawa(@shinyorke) • Pythonista/Agile Software Development/Baseball Analyst • visasQ(ϏβεΫ) Python Engineer/Scrum Master • ๺ւಓ೔ຊϋϜϑΝΠλʔζ/Oakland Athletics • ιχʔɾάϨΠ(OAK)ͷαΠϠϯά৆ड৆  &Ԭ޿ւ(೔ϋϜ)ͷελϝϯୣऔΛ৴͍ͯ͡·͢.

Slide 3

Slide 3 text

ࠓγʔζϯݟͲ͜Ζ ݟͲ͜Ζ ੈؒͷ෩ை த઒ͷݟղ ༏উνʔϜ ɾιϑτόϯΫ ɾϠΫϧτ ɾ೔ϋϜ ɾڊਓPS޿ౡ τϦϓϧεϦʔ ɾ༄ా༔ذ ೥࿈ଓ ɾࢁా఩ਓ ೥࿈ଓ ࢁా఩ਓ ೥࿈ଓ Ϊʔλ͸ࡾףͲ͏ͧ ΰʔϧσϯάϥϒ৆ ɾ༄ా༔ذ $' ɾௗ୩ܟ 44 ɾೋਓڞ຅ऩ ɾγϣʔτ͸୭͕ʁ ۙ౻݈հ ೔ϋϜ ɾׂຊ͍͚ΔͰʂ ɾࢦ໊ଧऀPSϥΠτ ۙ౻ ࢦcӈcัcࡾc༡ ˠॅॴෆఆʹͳΔ

Slide 4

Slide 4 text

Starting Member • ໺ٿHack!2015೥ৼΓฦΓ • MLBҰٿ଎ใσʔλͱ໺ٿHack • MLBҰٿ଎ใσʔλΛPythonͰHackͯ͠ΈΔ  ʙpitchpxͱJupyter + pandas + matplotlibʙ • ར༻ྫʙؠ۾ٱࢤϊʔώοτϊʔϥϯ • ݁ͼʙࠓޙͷ໺ٿHack(PyCon JP 2016ʹ޲͚ͯ) • ʲΦϚέʳ2016೥ϓϩ໺ٿେ༧૝

Slide 5

Slide 5 text

໺ٿHack!1.0(PyCon JP 2015) • MLBͷࢼ߹͝ͱͷଧ੮σʔλΛHack! • ࢄาʢ࢛ٿʣͷ਺ʢΠονVSϘοτʣ • ϐονϟʔͷ݄ผউͪ੕ʢδϣϯɾϨελʔʣ • ຖ೔ຖࢼ߹ͷσʔλΛऔಘ&෼ੳ • ΞμϜɾμϯ཰ʢଧऀʣ • ඃΞμϜɾμϯ཰ʢ౤खʣ • ৄ͘͠͸εϥΠυΛޚཡ͍ͩ͘͞ or ʮ໺ٿ PythonʯͰάάΖ͏

Slide 6

Slide 6 text

໺ٿHack!ʙPythonΛ༻͍ͨσʔλ෼ੳͱՄࢹԽ PyCon JP 2015ൃදࢿྉ http://www.slideshare.net/shinyorke/hackpython-pyconjp

Slide 7

Slide 7 text

໺ٿHack!ʙPythonΛ༻͍ͨσʔλ෼ੳͱՄࢹԽ PyCon JP 2015ൃදࢿྉ http://www.slideshare.net/shinyorke/hackpython-pyconjp ೥ͷωλ

Slide 8

Slide 8 text

໺ٿHack!ʙPythonΛ༻͍ͨσʔλ෼ੳͱՄࢹԽ PyCon JP 2015ൃදࢿྉ http://www.slideshare.net/shinyorke/hackpython-pyconjp Ұٿ଎ใ΍Γ͍ͨϯΰ ˠ೥ͷςʔϚʂ

Slide 9

Slide 9 text

໺ٿHack!ͱҰٿ଎ใ • ࢼ߹ɾଧ੮ͷ݁Ռetc…είΞͰଌΕΔωλ͸΍Γ੾ͬͨײ͋Δ • બखͷނোɾෆௐʢ޷ௐʣ͸είΞͰଌΕͳ͍ˠ΍Γ͍ͨ • ౤खͳΒٿ଎ɾίϯτϩʔϧɾϘʔϧͷճస਺ɺ  ໺ख͸कඋൣғ(଍)ɾεΠϯάεϐʔυͰଌΕΔͷͰ͸ʂʁ • Ұٿ଎ใͷσʔλ͕͋ΕͰ͖ͦ͏…͋ͬͨʂʂʂ • ࢼ͠ʹ΍ͬͯΈΑ͏ʂʂʂˡࠓίί

Slide 10

Slide 10 text

MLB at BATʙMLBҰٿ଎ใ • MLB࣮گҰٿ଎ใαʔϏε • PCαΠτɾεϚϗΞϓϦɾApple TVͳͲ • MLB.TVͱ߹Θͤͯܖ໿Ͱ࣮گಈը΋ݟΒΕΔ • σʔλ͕ͱʹ͔͘ॆ࣮

Slide 11

Slide 11 text

Analyzing Baseball Data with R • MLBͷΦʔϓϯσʔλʮRetrosheetʯ,  MLB at BAT଎ใσʔλΛ༻͍ͨσʔλ෼ੳɾՄࢹ Խʹ͍ͭͯॻ͔Ε͍ͯΔॻ੶ʢӳޠʣ • RݴޠΛ࢖ͬͨ෼ੳͱՄࢹԽͷωλ͕ϝΠϯ • ʮpitchRxʯͱ͍͏ɺRݴޠͷϥΠϒϥϦΛ༻͍ͯ  at BATσʔλΛऔಘ&ՄࢹԽ

Slide 12

Slide 12 text

“ʮpitchRxʯͱ͍͏ɺ  RݴޠͷϥΠϒϥϦΛ༻͍ͯ  at BATσʔλΛऔಘ&ՄࢹԽ”

Slide 13

Slide 13 text

ʁʁʁʮPythonͰ΍Γ͍ͨΜ͡Όʂʯ ※RΛͲ͏͜͏ݴ͏ͱ͔ͦΜͳҙਤ͸(ry

Slide 14

Slide 14 text

pitchpx - Getting MLB dataset • MLB at BATͷҰٿ଎ใσʔλΛऔಘ&εΫϨΠϐϯάͯ͠ CSVσʔληοτʹམͱ͢PythonϥΠϒϥϦ. • pitchRx(R)ͳͲΛࢀߟʹࢲ͕։ൃ͠·ͨ͠. • ίϚϯυϥΠϯπʔϧͰ͢. • Python 3.3.xҎ্ઐ༻ˡڧ͍ͩ͜ΘΓ • PyPIͰެ։͍ͯ͠·͢ʂʂʂʢ୭Ͱ΋࢖͑Δʣ

Slide 15

Slide 15 text

࢖͍ํ $ # Python 3.3Ҏ্(ਪ঑Python 3.4Ҏ্)͕ಈ͘؀ڥͰ΍ͬͯͶ $ pip install pitchpx $ # ྫɿ2015/8/1-8/12·Ͱͷࢼ߹݁ՌΛऔಘ͢Δ $ pitchpx -s 20150801 -e 20150812 -o .

Slide 16

Slide 16 text

ʲྫʳؠ۾ϊʔώοτϊʔϥϯ • ϚϦφʔζ-ΦϦΦʔϧζͷࢼ߹(2015/8/12)ʹͯɺ  ϊʔώοτϊʔϥϯΛܾΊͨؠ۾ٱࢤ౤खͷ౤ٿΛ෼ੳ • ٿ଎ɺϘʔϧͷճసɺετϥΠΫκʔϯɺ܏޲etc… • pitchpxͰऔಘͨ͠σʔλΛpandasͱ matplotlib(&seaborn)Ͱલॲཧ&ՄࢹԽ • ؀ڥ͸Jupyter notebook(Python 3.5.1)

Slide 17

Slide 17 text

σϞ (লུ)

Slide 18

Slide 18 text

ৄ͘͠͸QiitaͰʂʂʂ ؠ۾ٱࢤ(SEA)ͷφΠεϐονϯάΛPythonͰՄࢹԽ http://qiita.com/shinyorke/items/2c2e2c3976fc2d1ed051

Slide 19

Slide 19 text

݁ͼʙ2016೥ͷ໺ٿHack! • ͦΒʢࠓ೔͸౤ٿσʔλͷՄࢹԽ͔ͩΒʣ  ͦ͏ʢͭ͗͸कඋσʔλͷՄࢹԽʹʣ  Αɹʢܾ·͍ͬͯΔ͡Όͳ͍͔ʣ • PyCon JP 2016(9/21,22)͸ɺ  ʮAnalyzing Baseball Data With Pythonʯ  ͱ͔ͦΜͳλΠτϧͰ΋ͬͱ໘ന͍࿩͕Ͱ͖Δϋζ. • ຊ೔ެ։ͨ͠ωλ͸ੋඇ༡ΜͰΈͯʂ  ˠػցֶशͷ୊ࡐͱ͔ʹΠέΔΜ͡Όͳ͍ʁ

Slide 20

Slide 20 text

ʮҰٿ଎ใσʔλͷϥΠηϯε͸ʁେৎ෉ͳͷʁʯ ※Ұ൪͋Γͦ͏ͳ࣭໰

Slide 21

Slide 21 text

౴ɿ(ݸਓར༻ఔ౓ͳΒ)OK ʲެࣜʳ http://gd2.mlb.com/components/copyright.txt ʲ༁&ղઆʳ http://qiita.com/shinyorke/items/566f1b7e7687492a0c7f

Slide 22

Slide 22 text

ήʔϜηοτʂʂʂ ͝ਗ਼ௌ͋Γ͕ͱ͏͍͟͝·ͨ͠. Shinichi Nakagawa(Twitter/Facebook/hatena:@shinyorke)