tidyverse tutorial 1

tidyverse 超入門
講義用

Transcript

1. σʔλղੳͱલॲཧ .
   ࠇ໦༟ୋ
   !FEUVTBDKQ
2. ໨࣍  σʔλղੳ
   ͱ͸
    3
   ͱ
   34UVEJP
    3
   ͷجຊ
    ϞμϯͳσʔλϑϨʔϜૢ࡞ !2
3. ຊ೔࢖༻͢Δσʔλ TUBSXBST
   w ελʔ΢Υʔζͷొ৔ਓ෺ʹؔ͢Δσʔλ
   IUUQTXBQJDP
   qJHIUT
   w ೥ʹ
   -("
   +',
   &83
   Λग़ൃͨ͢͠΂ͯͷϑϥΠτͷఆࠁσʔλ
   XFBUIFS
   w -("
   +',
   &83
   ͷఱީ΍෩ͷ৘ใ
   ࣌ؒ͝ͱ
   BJSMJOFT
   w ߤۭձࣾͷςʔϒϧ !3
4. σʔλղੳͱ͸
5. σʔλղੳͱ͸ !5 6OEFSTUBOE *NQPSU 0VUQVU
   
   4IBSF
6. σʔλղੳͱ͸ !6 *NQPSU 0VUQVU
   
   4IBSF 7JTVBMJ[F .PEFMMJOH *OUFSQSFU 5SBOTGPSN
7. 5PPMT ͨ͘͞Μ͋Δ
   w ి୎
   w .JDSPTPGU
   &YDFM
   w *#.
   4144
   w 4"4
   w 1ZUIPO
   w 3 !7
8. 5PPMT ޷͖ʹબ΂͹ྑ͍
   w ޷Έ
   ྲྀߦΓ
   څྉ
   ਎ۙʹಘҙͳਓ͕͍Δ
   FUD !8 #JH
   %BUB ػցֶश ౷ܭղੳ ϋϯυϦϯά 42- ˕ ✕ ✕ ˓ 3 ˚ ˓ ˕ ˕ 1ZUIPO ˚ ˕ ˓ ˓ &YDFM ✕ ✕ ✕ ✕ ి୎ ✕ ✕ ✕ ✕
9. 3
   ͱ
   34UVEJP
10. 3 ಘҙͳͷͰ3Λ঺հ͠·͢
    
    w ΠϯλϓϦλݴޠ
    㲗ίϯύΠϥݴޠ
    ௕ॴ
    w 044
    w ༷ʑͳ
    ౷ܭղੳɾػցֶश
    ϥΠϒϥϦ
    w 34UVEJP
    ࣾͷଘࡏ
    ୹ॴ
    w ஗͍
    ಺෦͕$ ͳͲͰॻ͔Ε͍ͯΔͱ଎͍ !10
11. 34UVEJP
    Λ࢖͓͏ w 3ʹಛԽͨ͠౷߹։ൃ؀ڥ
    *%&
    w ΤσΟλͱͯ͠༏लͳ͚ͩͰͳ͘
    ༷ʑͳ֦ுػೳ
    w ๛෋ͳγϣʔτΧοτ
    w (JU
    ͱͷ࿈ܞ
    w ແྉ !11
12. 34UVEJP
    ͷը໘  ΤσΟλ
     ίϯιʔϧ
     ΦϒδΣΫτ
    ཤྺ
    ͳͲ
     ϓϩοτ
    ϔϧϓ
    ͳͲ !12 ᶃ ᶄ ᶅ ᶆ
13. ىಈ͠Α͏ w ىಈ
    w ഑෍ͨ͠
    &YFSDJTF31SPK
    Λ্ཱͪ͛Δ
    w 3
    Λಈ͔ͯ͠ΈΑ͏
    w ίϯιʔϧͰ
    plot(iris)
    w ύοέʔδΛΠϯετʔϧͯ͠ΈΑ͏
    ͕͔͔࣌ؒΔͷͰ஫ҙ
    • install.packages(“tidyverse”) w εΫϦϓτΛ࡞ͬͯΈΑ͏ w 'JMF
    
    
    
    /FX
    'JMF
    
    
    
    3
    4DSJQU !13
14. 31SPKFDU
    ͱ͸ w ϓϩδΣΫτ୯ҐͰ෼ੳΛ؅ཧ͢Δ͘͠Έ
    w Ұ࿈ͷղੳͰඞཁʹͳΔϑΝΠϧ܈Λ·ͱΊͯѻ͏
    w ࡞ۀσΟϨΫτϦ͕ͦ͜ʹͳΔ
    w ଞਓͱͷڞ༗͕͠΍͍͢
    w 'JMF
    
    
    
    /FX
    1SPKFDU
    
    Ͱ࡞੒ !14
15. ؀ڥΛઃఆ͠Α͏ w 5PPMT
    
    
    
    (MPCBM
    0QUJPOT
    
    ͔Β؀ڥઃఆ
    w (FOFSBMɿνΣοΫΛશͯ֎ͦ͏
    w 3FTUPSF
    QSFWJPVTMZ
    PQFO
    TPVSDF
    EPDVNFOU
    BU
    TUBSU
    VQ
    
 ͸࢒͓͍ͯͯ͠ྑ͍
    w $PEF
    
    
    
    4BWJOHɿΤϯίʔσΟϯάΛ
    65'
    ʹ
    w "QQFBSBODFɿ޷͖ͳݟͨ໨ʹઃఆ͠Α͏ !15
16. 3ͷجຊ
17. جຊ w GPSจ
    JGจ
    ؔ਺
    ͦͷଞʜ
    w ଞͷݴޠͱ͋·ΓมΘΒͳ͍ͷͰ
    ͻͭΑ͏ʹͳͬͨΒௐ΂ ͍ͯͩ͘͞
    
    w ഑ྻͷࢀর͸͸͡·Γʂʂ !17
18. 3
    Ͱͷσʔλܕ Ұͭͷ஋
    w -PHJDBM
    #PPMFBO
    w *OUFHFS
    w %PVCMF
    w $PNQMFY
    w $IBSBDUFS
    w 'BDUPS
    w FUDʜ !18 ෳ਺ͷ஋
    w "UPNJD
    7FDUPS
    w .BUSJY
    w %BUB
    'SBNF
    w -JTU
    w FUDʜ
19. "UPNJD
    7FDUPS w ̍࣍ݩ഑ྻ
    w ཁૉ͸શͯಉ͡ܕ
    શͯJOUFHFS
    શͯDIBSBDUFS
    ͳͲ
    w ཁૉͷશͯΛ·ͱΊͯॲཧͰ͖Δʢ࢛ଇԋࢉͱ͔ʣ
    w c()
    Ͱ࡞Δ !19
20. "UPNJD
    7FDUPS !20 drink <- c(“beer”, “sake”, “whisky”) # ୅ೖ drink # ΦϒδΣΫτͷݺͼग़͠ price <- c(480, 700, 850) # ਺஋ܕϕΫτϧ favorite <- c(TRUE, TRUE, TRUE) # ࿦ཧܕϕΫτϧ
21. .BUSJY w ̎࣍ݩ഑ྻ
    w ཁૉ͸શͯಉ͡ܕ
    w ߦྻԋࢉ͕Ͱ͖Δʢ಺ੵͱ͔ʣ
    w ཁૉͷશͯΛ·ͱΊͯॲཧͰ͖Δʢ࢛ଇԋࢉͱ͔ʣ
    w matrix()
    Ͱ࡞Δ !21
22. -JTU w ̍࣍ݩ഑ྻ
    w ཁૉ͸ͳΜͰ΋͍͍
    7FDUPS
    ͷ֦ு
    w list()
    Ͱ࡞Δ
    w ࣗ༝౓͕ΊͬͪΌߴ͍ !22
23. %BUB
    'SBNF ͓ͳ͡Έͷ࢛͍֯ςʔϒϧ
    • ֤ྻ͸ಉ͡௕͞ͷ Atomic vector w data.frame()Ͱ࡞Δ
    w data.frame(drink = drink, 
 price = price, 
 favorite = favorite)
    w ༷ʑͳύοέʔδ͕%BUB
    'SBNFΛத৺ʹ࡞ΒΕ͍ͯΔ !23
24. %BUB
    'SBNF
    ʹ৮Ζ͏ starwars <- read.csv("data/starwars.csv", stringsAsFactors = FALSE, fileEncoding = “UTF-8”) head(starwars) # ઌ಄֬ೝ tail(starwars) # ຤ඌ֬ೝ summary(starwars) # هड़౷ܭྔ str(starwars) # ֤ྻͷܕ֬ೝ !24
25. ϞμϯͳσʔλϑϨʔϜૢ࡞
26. ʮ5PPM
    ʹਫ਼௨͢Δʯͱ͍͏͜ͱ ྉཧʹྫ͑Δͱ
    w แஸ΍ίϯϩΛ࢖͑ΔΑ͏ʹͳΖ͏
    w Ϩγϐ͸దٓݟΕ͹Α͍
    ʮര଎ͰσʔλΛѻ͑Δʯͱ͍͏͜ͱ
    w ๲େͳࢼߦࡨޡΛ܁ΓฦͤΔ
    w ࣌ؒ͸༗ݶ
    ຊ࣭తͰͳ͍࡞ۀ͸ͬ͞͞ͱऴΘΒͤͯ
    
 ҿΈʹग़͔͚Α͏
    ݚڀ͠Α͏ !26
27. 5JEZWFSTF ֓೦
    w 3
    Ͱͷ༷ʑͳૢ࡞
    *NQPSU
    &YQPSU
    5SBOTGPSN
    7JTVBMJ[BUJPO
    FUD
    ͕
 ౷ҰతͳΠϯλʔϑΣʔεͰग़དྷͨΒ
    ૉఢͩΑͶ
    ύοέʔδ
    w ্هΛ࣮ݱ͢ΔͨΊͷύοέʔδ܈
    w install.packages(“tidyverse”) ͰΠϯετʔϧ
    w )BEMFZ
    8JDLIBN
    34UVEJPࣾ
    ͕த৺ͱͳΓ։ൃ
    w ௒ศར !27 ˞
    )BEMFZ
    8JDLIBN
    3
    քͷਆ
28. 5JEZWFSTF !28
29. 5JEZWFSTF !29 ಛʹ͜ΕΒ
30. library(tidyverse)
31. %BUB
    'SBNF
    ͷجຊૢ࡞
    EQMZS w ม਺
    ྻ
    ͷநग़
    w ؍ଌ
    ߦ
    ͷநग़
    w ؍ଌ
    ߦ
    ͷฒͼସ͑
    w ৽ͨͳม਺
    ྻ
    ͷ࡞੒
    w ूܭ
    w άϧʔϓԽ !31 • select() • filter() • arrange() • mutate() • summarise() • group_by()
32. ࢖͍ํ w ୈҾ਺ʹ͸σʔλϑϨʔϜΛ༩͑Δ
    w ୈҾ਺Ҏ߱Ͱ͸ྻ໊ΛΫΦʔςʔγϣϯແ͠Ͱ༩͑Δ
    w ໭Γ஋͸৽ͨͳσʔλϑϨʔϜ !32
33. ΍ͬͯΈΑ͏ select(starwars, name, gender, species) filter(starwars, species == "Human", height <= 170) mutate(starwars, BMI = mass / (height/100)^2) arrange(starwars, gender, height) summarise(starwars, mean_mass = mean(mass, na.rm = TRUE), mean_height = mean(height, na.rm = TRUE)) grouped <- group_by(starwars, species) summarise(grouped, mean_mass = mean(mass, na.rm = TRUE), mean_height = mean(height, na.rm = TRUE), count = n()) !33
34. %>%
35. ύΠϓԋࢉࢠ
    %>% X %>% f X %>% f(y) X %>% f %>% g X %>% f(y, .) !35 f(X) f(X, y) g(f(X)) f(y, X) લͷؔ਺ͷग़ྗΛ࣍ͷؔ਺ͷୈҾ਺ʹΘͨ͢΋ͷ
    $NE
    
    4IJGU
    
    N
    $USM
    
    4IJGU
    
    N
    Ͱೖྗ
36. ෳ਺ͷॲཧΛ͢Δ৔߹ df1 <- filter(starwars, species == "Human") df2 <- mutate(d1, BMI = mass / (height/100)^2) df3 <- group_by(df2, gender) df4 <- summarise(df3, mean_BMI = mean(BMI, na.rm=TRUE), min_BMI = min(BMI, na.rm=TRUE), max_BMI = max(BMI, na.rm=TRUE) !36 # A tibble: 2 x 4 gender mean_BMI min_BMI max_BMI <chr> <dbl> <dbl> <dbl> 1 female 22.0 16.5 27.5 2 male 26.0 21.5 37.9
37. ෳ਺ͷॲཧΛ͢Δ৔߹ starwars %>% filter(species == "Human") %>% mutate(BMI = mass / (height/100)^2) %>% group_by(gender) %>% summarise(mean_BMI = mean(BMI, na.rm=TRUE), min_BMI = min(BMI, na.rm=TRUE), max_BMI = max(BMI, na.rm=TRUE)) %>%
    Λ࢖͏͜ͱͰ
    ୭ʹͰ΋ಡΈ΍͍͢ίʔυʹʂʂ !37
38. ࿅श໰୊  ਎௕ͷ࠷΋௿͍உੑΛ֬ೝͤΑ
     #.*ͷ΋ͬͱ΋ߴ͍ొ৔ਓ෺͸୭͔
     ฏۉ਎௕͕࠷΋ߴ͍छ଒͸Կ͔

    ݕࡧͯͦ͠ͷ࢟Λ͔֬ΊΑ͏ !38
39. ࣍ճ·Ͱͷ՝୊
40. ՝୊ 1. ൃۭߓ (origin) ͝ͱͷඈߦػศͷ਺, ඈߦڑ཭ͷฏۉ, ग़ൃ࣌ࠁ஗Ԇͷ ฏۉΛٻΊΑ 2. ೔෇͝ͱͷग़ൃ࣌ࠁ஗ԆͷฏۉΛٻΊΑ 3. ೔෇͝ͱʹܽߤʹͳͬͨศͷ਺Λௐ΂Α
 ʢܽߤͩͱdep_delayͱarr_delay͕NAʹͳΔʣ 4. ࣮ࡍʹඈΜͩศͷඈߦڑ཭ͷඪ४ภࠩΛग़ൃۭߓ͝ͱʹٻΊ, ঢॱʹ ฒ΂Α 5. ೔෇͝ͱʹ, ۭߓLGA͔Β࠷ॳʹඈΜͩศͱ࠷ޙʹඈΜͩศΛநग़ͤΑ 6. ࣮ࡍʹඈΜͩศͷ͏ͪ, ఆࠁ௨Γग़ൃͨ͠ศͷׂ߹Λௐ΂Α ώϯτɿdplyrͷؔ਺ͷதͰ n() Λ࢖͏ͱߦ਺Λࢉग़Ͱ͖Δ !40