Verbatim Practive

if("rJava" %in% installed.packages("rJava") == FALSE)install.packages("rJava")
library(rJava)
if("memoise" %in% installed.packages("memoise") == FALSE)install.packages("memoise")
library(memoise)
if("stringr" %in% installed.packages("stringr") == FALSE)install.packages("stringr")
library(stringr)
if("KoNLP" %in% installed.packages("KoNLP") == FALSE)install.packages("KoNLP")
library(KoNLP)
if("wordcloud" %in% installed.packages("wordcloud") == FALSE)install.packages("wordcloud")
library(wordcloud)
if("RColorBrewer" %in% installed.packages("RColorBrewer") == FALSE)install.packages("RColorBrewer")
library(RColorBrewer)
if("dplyr" %in% installed.packages("dplyr") == FALSE)install.packages("dplyr")
library("dplyr")

install.packages("readxl")
library("readxl")

setwd('C:\\Users\\Administrator\\rlang_weekend2\\Data_R_181013')
verbatim  <- read_excel("Verbatim.xlsx", 1)
# 만약 sheet가 여러개이면 숫자만, 2, 3, 4로 변경가능
head(verbatim)

useSejongDic()
KoNLP::buildDictionary(
  ext_dic = c('sejong','woorimalsam')
)

verbatim <- str_replace_all(verbatim,"\\W"," ")
head(verbatim)

nouns <- sapply(
                verbatim,
                extractNoun,
                USE.NAMES = F
)

nouns <- extractNoun(verbatim)
wordcount <- table(unlist(nouns))
df_word <- as.data.frame(wordcount,stringAsFactor = F)
df_word <- dplyr::rename(df_word,
                         word = Var1,
                         freq = Freq)
# head(df_word)
df_word <- filter(df_word, nchar(word)>=2)

top20 <- df_word %>% 
  arranage(desc(freq)) %>% 
  head(20)