daru チートシート

このチートシートは
datacamp.com 作成の Python For Data Science Cheat Sheet を Ruby で模したものです．
模したというものの完全に真似ることはできていません．また適宜変更も行っています．

daru の読み込みは下記のように行います．

[1] pry(main)> require "daru"

Install the reportbuilder gem version ~>1.4 for using reportbuilder functions.

Install the spreadsheet gem version ~>1.1.1 for using spreadsheet functions.
=> true

daru のデータ構造

Vector

VectorはpandasのSeriesに相当するものです．
1次元のlabeled arrayです．

[2] pry(main)> s = Daru::Vector.new([3, -5, 7, 4], index: [:a, :b, :c, :d])
=> #<Daru::Vector(4)>
   a   3
   b  -5
   c   7
   d   4

DataFrame

2次元のlabeled データ構造です．

[3] pry(main)> data = {'Country': ['Belgium', 'India', 'Brazil'], 'Capital': ['Brussels', 'New Delhi', 'Brasília'], 'Population': [11190846, 1303171035, 207847528]}
=> {:Country=>["Belgium", "India", "Brazil"], :Capital=>["Brussels", "New Delhi", "Brasília"], :Population=>[11190846, 1303171035, 207847528]}
[4] pry(main)> df = Daru::DataFrame.new(data)
=> #<Daru::DataFrame(3x3)>
               Capital    Country Population
          0   Brussels    Belgium   11190846
          1  New Delhi      India 1303171035
          2   Brasília     Brazil  207847528

選択

取得

一要素の取得

[5] pry(main)> s[:b]
=> -5

データフレームのサブセットの取得

[6] pry(main)> df.row[1..2]
=> #<Daru::DataFrame(2x3)>
               Capital    Country Population
          1  New Delhi      India 1303171035
          2   Brasília     Brazil  207847528

選択、ブーリアンインデックス そして 代入

ポジションによるもの (行と列のindexの順が通例とは逆であることに注意)

[7] pry(main)> df[1][0]
=> "Belgium"

ラベルによるもの

[8] pry(main)> df[:Capital][0]
=> "Brussels"

ブーリアンインデックス

[9] pry(main)> s.where(s.gt(1))
=> #<Daru::Vector(3)>
   a   3
   c   7
   d   4
[10] pry(main)> s.where(s.lt(-1) | s.gt(2))
=> #<Daru::Vector(4)>
   a   3
   b  -5
   c   7
   d   4
[11] pry(main)> df.where(df[:Population].gt(1200000000))
=> #<Daru::DataFrame(1x3)>
               Capital    Country Population
          1  New Delhi      India 1303171035

代入

[12] pry(main)> s[:a]=6
=> 6
[13] pry(main)> s
=> #<Daru::Vector(4)>
   a   6
   b  -5
   c   7
   d   4

ソートとランク

[14] pry(main)> df.sort([:Country])
=> #<Daru::DataFrame(3x3)>
               Capital    Country Population
          0   Brussels    Belgium   11190846
          2   Brasília     Brazil  207847528
          1  New Delhi      India 1303171035

VectorとDataFrameの情報の取得

基本情報

[15] pry(main)> df.shape
=> [3, 3]
[16] pry(main)> df.index
=> #<Daru::Index(3): {0, 1, 2}>
[17] pry(main)> df.vectors
=> #<Daru::Index(3): {Capital, Country, Population}>
[18] pry(main)> df.count
=> #<Daru::Vector(1)>
                 count
 Population          3

サマリー

[19] pry(main)> df.sum
=> #<Daru::Vector(1)>
                   sum
 Population 1522209409
[20] pry(main)> df.cumsum
=> #<Daru::DataFrame(3x1)>
            Population
          0   11190846
          1 1314361881
          2 1522209409
[21] pry(main)> df.min
=> #<Daru::Vector(1)>
                   min
 Population   11190846
[22] pry(main)> df.max
=> #<Daru::Vector(1)>
                   max
 Population 1303171035
[23] pry(main)> df.describe
=> #<Daru::DataFrame(5x1)>
            Population
      count          3
       mean 507403136.
        std 696134594.
        min   11190846
        max 1303171035
[24] pry(main)> df.mean
=> #<Daru::Vector(1)>
                                mean
        Population 507403136.3333333
[25] pry(main)> df.median
=> #<Daru::Vector(1)>
                median
 Population  207847528

I/O

CSVの読み書き

事前に csv file を入手します．

wget https://raw.githubusercontent.com/fivethirtyeight/data/master/airline-safety/airline-safety.csv`

[26] pry(main)> df = Daru::DataFrame.from_csv("airline-safety.csv")
=> #<Daru::DataFrame(56x8)>
               airline avail_seat incidents_ fatal_acci fatalities incidents_ fatal_acci fatalities
          0 Aer Lingus  320906734          2          0          0          0          0          0
          1  Aeroflot* 1197672318         76         14        128          6          1         88
          2 Aerolineas  385803648          6          0          0          1          0          0
          3 Aeromexico  596871813          3          1         64          5          0          0
          4 Air Canada 1865253802          2          0          0          2          0          0
          5 Air France 3004002661         14          4         79          6          2        337
          6 Air India*  869253552          2          1        329          4          1        158
          7 Air New Ze  710174817          3          0          0          5          1          7
          8 Alaska Air  965346773          5          0          0          5          1         88
          9   Alitalia  698012498          7          2         50          4          0          0
         10 All Nippon 1841234177          3          1          1          7          0          0
         11  American* 5228357340         21          5        101         17          3        416
         12 Austrian A  358239823          1          0          0          1          0          0
         13    Avianca  396922563          5          3        323          0          0          0
         14 British Ai 3179760952          4          0          0          6          0          0
        ...        ...        ...        ...        ...        ...        ...        ...        ...
[27] pry(main)> df.write_csv("aaa.csv")
=> nil
[28] pry(main)> .head aaa.csv
airline,avail_seat_km_per_week,incidents_85_99,fatal_accidents_85_99,fatalities_85_99,incidents_00_14,fatal_accidents_00_14,fatalities_00_14
Aer Lingus,320906734,2,0,0,0,0,0
Aeroflot*,1197672318,76,14,128,6,1,88
Aerolineas Argentinas,385803648,6,0,0,1,0,0
Aeromexico*,596871813,3,1,64,5,0,0
Air Canada,1865253802,2,0,0,2,0,0
Air France,3004002661,14,4,79,6,2,337
Air India*,869253552,2,1,329,4,1,158
Air New Zealand*,710174817,3,0,0,5,1,7
Alaska Airlines*,965346773,5,0,0,5,1,88