Slide 1

Slide 1 text

Tokyo.R ver.62 LT ʮWord Predictionʯ 2017/06/24(sat) @ࣚཹ

Slide 2

Slide 2 text

ࣗݾ঺հ ໊લɿҏ౻ ప࿠(@tetsuroito) ࢓ࣄɿFinTechܥ झຯɿञɺαοΧʔ؍ઓɺεΩʔ ݴޠɿSQL΍Rݴޠɻ࠷ۙ͸Python΍ͬͯΔ ࠷ۙ͸෼ੳͱ͔͋Μ·Γ͍ͯ͠ͳ͍

Slide 3

Slide 3 text

એ఻ɿ࿈ࡌ΍ͬͯ·͢ ιʔγϟϧ֦ࢄͷఆྔσʔλ͕ࢲͷϞνϕͰ͢

Slide 4

Slide 4 text

ࠓ೔ͷLTͷ͖͔͚ͬ ;ͱ31VCTͷΤϯτϦΛݟ͍ͯͨΒɺ Կ΍Β໘നͦ͏ͳ΋ͷΛൃݟͨ͠ͷͰɺ ࠓ೔͸͜Εͷ࿩Ͱ͢ IUUQSQVCTDPN.BMPSFBO

Slide 5

Slide 5 text

എܠͱϞνϕʔγϣϯ ɾܞଳ୺຤͸ΩʔϘʔυϨΠΞ΢τʹ͸খ͍͞ ɾ୹ॖͨ͠ϫʔυΛଧͪࠐΉ͍͔ͭ͘ͷख๏͕͋Δ T9 (Text on 9keys):ΨϥέʔϘλϯΈ͍ͨͷ Sliding:εϚϑΥͷΩʔϘʔυ ༧ଌม׵ ͜ͷ1+͸ೖྗ͞ΕͨϑϨʔζʹ࠷΋͋Γͦ͏ͳޠΛ ༧ଌͯ͠ఏࣔ͢Δͱ͍͏΋ͷ

Slide 6

Slide 6 text

ߏ੒ Capstone Dataset RͰ࣮ݱ ख๏ɿTMɺQuantedaɺtext2vec DBɿSqlite using RSQlite εϐʔυͱγϯϓϧ͞Ͱ্هͷબఆ

Slide 7

Slide 7 text

σʔλϞσϧ ετοϓϫʔυͳ͠ N-GramΛར༻Ͱ(2-Gram͔Β7-Gram) ༧ଌม׵

Slide 8

Slide 8 text

N-Gram n-1ޠΛจ຺ͱͯ࣍͠ͷޠΛ༧ଌ จࣈn-gram ୯ޠn-gram class n-gramͳͲ ࣗવݴޠॲཧʹ͓͚ΔҰൠతͳݴޠϞσϧͰ͢

Slide 9

Slide 9 text

݁Ռ͸Shiny Appʹ IUUQTNBMPSFBOTIJOZBQQTJP8PSE1SFEJDUJPO

Slide 10

Slide 10 text

݁࿦ ɾ՝୊ ɹ5.ύοέʔδͷେ͖͍σʔληοτͷύϑΥʔϚϯε ɹΠϯϑϧΤϯβʹ͔͔ͬͯ࣌ؒͱΒΕͪΌͬͨ ɾֶͼ ɹXPSLJUFSBUJWF ɹ΋ͬͱίʔυ΍σʔλখ͘͞Ͱ͖Δ͔΋ ɾࠓޙͷൃలʹΉ͚ͯ ɹҧ͏σʔλͰࢼ͍ͨ͠ ɹ4LJQ(SBNΛ࢖͏ ɹ,OFTFS/FZ΍,BU[`TCBDLP⒎ͰεϜʔδϯά

Slide 11

Slide 11 text

͝੩ௌ͋Γ͕ͱ͏ ͍͟͝·ͨ͠ʂ