In rommelnc/hudar: High-Level Unstructured Data Analysis

Política

#install_github('rCharts', 'ramnathv', ref = 'dev')
require(streamR)
library(rCharts)
# rCharts::open_notebook()
# tweets <- parseTweets("../../data/politics-tweets-2014-05-14.json", simplify = TRUE)
# tweets <- rbind(tweets, parseTweets("../../data/politics-tweets-2014-05-15-antes de travar.json", simplify = TRUE))
# tweets <- rbind(tweets, parseTweets("../../data/politics-tweets-2014-05-15.json", simplify = TRUE))
# save(tweets, file='../../data/politics-tweets-2014-05-14 e 15.Rda')
load('../../data/politics-tweets-2014-05-16a19.Rda')

# eleicoes = tweets[grep("eleicao|eleição|eleicoes|eleições|Dilma Rousseff|Aécio Neves|Aecio Neves|Eduardo Campos|Marina Silva", tweets$text, ignore.case=TRUE),][1:500,]
eleicoes = tweets[1:100,]

eleicoes$text = gsub("[^[:graph:]]", " ",eleicoes$text) #remove caracteres inválidos

source('funcoesTwitter.R')

# Encoding(eleicoes$text) = 'ISO-8859-1'

lista = montarListaUsuarios(eleicoes)
tabelas = montarTabelasUsuarios(lista$listaDePara)

Os 10 tweets mais frequentes:

sort(table(eleicoes$text), decreasing=T)[1:10]

Before we begin charting, let us customize some settings for rCharts and knitr

require(rCharts)
options(RCHART_WIDTH = 600, RCHART_HEIGHT = 400)
knitr::opts_chunk$set(comment = NA, results = 'asis', tidy = F, message = F)

Gráficos:

maisFreq = sort(table(eleicoes$text), decreasing=T)[1:10]
tweets = paste(names(maisFreq), collapse="@!K#$%@")

df1 = as.data.frame(names(tabelas$maisFalou[1:10]))
df1$qtd = array(tabelas$maisFalou[1:10])
df1$tweets = tweets
names(df1) = c("usuario", "qtd", "tweets")
n1 = nPlot(x="usuario", y="qtd", data=df1, type="discreteBarChart")
n1$xAxis(rotateLabels = -20)
n1$xAxis(axisLabel = "Usuários") 
n1$yAxis(axisLabel = "Quantidade de Tweets")
n1$chart(margin = list(left = 90))
n1$chart(margin = list(bottom = 100))
# n1$templates$script <- "http://timelyportfolio.github.io/rCharts_nvd3_templates/chartWithTitle.html"
n1$set(title = "Os 10 usuários que mais falaram")


n1$chart(tooltipContent= "#! function(key, x, y, e){
  var tweets = e.point.tweets.split('@!K#$%@');
  var result = '';
  tweets.forEach(function(tweet) {
      result += tweet + '<br/>'
  });
  return result;
} !#")
n1$print('chart1')

barplot(tabelas$maisFalado[1:10], las=2)
title('Os 10 usuários mais falados')

barplot(tabelas$maisOcorrencias[1:10], las=2)
title('Os 10 usuários com mais ocorrências')

rommelnc/hudar documentation built on May 27, 2019, 1:49 p.m.