CÓDIGOS ÚTEIS NO R - Datassaurus

---

Visualização de dados - Datassaurus

library(ggplot2)
library(readr)
library(psych)

Attaching package: 㤼㸱psych㤼㸲

The following objects are masked from 㤼㸱package:ggplot2㤼㸲:

    %+%, alpha

#base de dados
datassaurus <- read_csv("Datasaurus_data.csv")

Parsed with column specification:
cols(
  x = col_double(),
  y = col_double()
)

datassaurus

# correlação
cor(datassaurus)

          x         y
x 1.0000000 0.1085411
y 0.1085411 1.0000000

# Estatísticas Descritivas
describe(datassaurus)

  vars   n  mean    sd median trimmed   mad   min    max  range skew kurtosis   se
x    1 142 57.30 39.42  53.72   53.94 16.16 22.31 479.49 457.18 8.63    89.75 3.31
y    2 142 47.83 26.94  46.03   46.90 30.79  2.95  99.49  96.54 0.25    -1.06 2.26

Existe algo de errado na Figura?

# Plot
g <- ggplot(datassaurus, aes(x, y))
g + geom_point() + 
    labs(title="Datassaurus", 
         subtitle="Vamos Visualizar os dados?",
         x="X",
         y="Y")

Qual é a observação discrepante?

outlier(datassaurus)

           1            2            3            4            5            6 
  3.41842761   3.31857472   3.22450563   2.91585947   2.60441986   2.28084791 
           7            8            9           10           11           12 
  1.88091772   1.76414743   1.66976851   1.55916722   1.37577756   1.17051550 
          13           14           15           16           17           18 
  0.98704506   0.82391774   0.67725695   0.53644185   0.42071661   0.34692171 
          19           20           21           22           23           24 
  0.21969721   0.16394920   0.14229959   0.12281826   0.10350097   0.09497209 
          25           26           27           28           29           30 
  0.14112902   0.28560402   0.47679752   0.62440241   0.07668278   0.01067438 
          31           32           33           34           35           36 
  0.01740176   0.16186611   0.07520201   0.04021454   0.24629409   0.41597326 
          37           38           39           40           41           42 
  0.53791957   0.67042809   0.66015386   0.59300440   0.39219378   0.27465296 
          43           44           45           46           47           48 
  0.24477475   0.62170444   0.74707838   0.86892165   0.92260810   0.90926318 
          49           50           51           52           53           54 
  0.80737304   0.73971532   0.53628326   0.38936196   0.24046230   0.14814570 
          55           56           57           58           59           60 
  0.07197236   0.04408743   0.01936323   0.02273950   0.04043941   0.07357690 
          61           62           63           64           65           66 
  0.14588202   0.32308868   0.46981920   0.69615340   1.57072347   1.95702174 
          67           68           69           70           71           72 
  2.04472151   1.99779922   2.05213077   2.03986799   1.72917365   1.53257833 
          73           74           75           76           77           78 
  1.39257481   1.41189086   1.34745676   0.76541283   0.86350143   0.82292234 
          79           80           81           82           83           84 
  3.71491290   3.63182664   3.40903339   2.96037762   2.59991890   2.09972136 
          85           86           87           88           89           90 
  1.81458162   1.45545724   1.18330613   0.94666817   0.74145692   0.54300282 
          91           92           93           94           95           96 
  0.47817537   0.50149300   0.55348690   0.62764432   0.74837738   0.99317408 
          97           98           99          100          101          102 
  1.20133042   1.48682761   0.07144631   0.13206155   0.29580468   0.53337485 
         103          104          105          106          107          108 
  0.77274506   0.85245926   1.14089840   1.06286793   1.04816031   1.15776227 
         109          110          111          112          113          114 
  1.20726608   1.18519670   1.22364960   1.20351448   1.19946812   1.26769916 
         115          116          117          118          119          120 
  1.40949712   1.63609095   1.50470818   1.49780613   1.50010477   1.57964982 
         121          122          123          124          125          126 
  1.71450521   1.87214503   1.99707642   2.42227118   2.02693487   1.95023663 
         127          128          129          130          131          132 
  1.97037636   1.95471615   2.01165998   2.06352464   2.16406047   2.24992580 
         133          134          135          136          137          138 
  2.37459676   2.29384935   2.70768154   2.98844946   0.83985802   0.82676438 
         139          140          141          142 
  0.85142214   3.31224065 115.05822598   3.04282355 

# excluir a linha 141º
datassaurus <- datassaurus[-141,]
g <- ggplot(datassaurus, aes(x, y))
g + geom_point() + 
    labs(title="Datassaurus", 
         subtitle="Vamos Visualizar os dados?",
         x="X",
         y="Y")