힙합 가사 텍스트 마이닝

패키지 준비

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("KoNLP" %in% installed.packages("KoNLP") == FALSE)install.packages("KoNLP")
library(KoNLP)
if("tm" %in% installed.packages("tm") == FALSE)install.packages("tm")
library(tm)
if("wordcloud" %in% installed.packages("wordcloud") == FALSE)install.packages("wordcloud")
library(wordcloud)
if("dplyr" %in% installed.packages("dplyr") == FALSE)install.packages("dplyr")
library(dplyr)
library(ggplot2)
if("stringr" %in% installed.packages("stringr") == FALSE)install.packages("stringr")
library(stringr)
if("RColorBrewer" %in% installed.packages("RColorBrewer") == FALSE)install.packages("RColorBrewer")
library(RColorBrewer)

KoNLP::useSejongDic()
getwd()

setp 1. 데이터를 로딩한다

txt <- readLines("hiphop.txt")
head(txt)

step 2. 특수문자를 제거

txt <- stringr::str_replace_all(txt, "\\W", " ") # 특수문자 제거, 한글만 남기는건 대문자 W
txt
txt <- stringr::str_replace_all(txt,"[^[:alpha:]]"," ") 

step 3. 명사만 추출

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

step 4. 단어별로 빈도표 작성

wordcount <- table(unlist(nouns))
df_word <- as.data.frame(wordcount, stringsAsFactors = F)

step 5. 변수명 수정

names(df_word)
df_word <- dplyr::rename(
  df_word,
  word = Var1,
  freq = Freq
)
df_word

step 6. 2글자 이상만 추출

df_word <- dplyr::filter(df_word, nchar(word)>=2)
df_word

step 7. 빈도순 정렬 후 상위 20단어만 추출

top_20 <- df_word %>% 
  dplyr::arrange(desc(freq)) %>% 
  head(20)
top_20

step 8. Word cloud 만들기

wordcloud::wordcloud(
  words = df_word$word,
  freq = df_word$freq,
  min.freq = 2,
  max.words = 200,
  random.order = F,
  rot.per = .1,
  scale = c(4, 0.3),
  colors = brewer.pal(8, "Dark2")
)