Commit b4ab99ac authored by amra669c's avatar amra669c
Browse files

Finalize PoDS assignment: cleaning, sentiment analysis, README

parent 4ef6be93

02_code/R/DataCleaning.R

+23 −14
Original line number Diff line number Diff line 
# Clear workspace

rm(list = ls())



# Set working directory

setwd("C:/Users/golna.GOLNAZ/Desktop/Hohenheim/Data Science/Assignment")

# -----------------------------

# Setup

# -----------------------------

# Set working directory where data and scripts are stored



# necessary packages

install.packages("hunspell")

install.packages("wordcloud2")

install.packages("tm")

# setwd()





# Required packages: hunspell, wordcloud2, tm, stringr



library(stringr)

library(hunspell)
@@ -19,7 +21,7 @@ library(tm) 


# Load the reviews

load("01_data/raw/gamereviews.RData")

View(game.rev)







# TASK 3: CLEANING
@@ -90,12 +92,8 @@ cat("Cleaned reviews:", nrow(reviews_final), "\n") 
cat("Saved to: 01_data/reviews_final.RData\n")



# Create a vector with all words in the reviews

all.words <- NULL

all.words <- unlist(str_split(reviews.final$review, " "))



for (i in 1:nrow(reviews.final)){

  tmp <- str_split(reviews.final$review[i], " ", simplify = TRUE)

  all.words <- c(all.words, tmp)

}



# sort the words in alphabetical order

all.words <- sort(all.words)
@@ -130,5 +128,16 @@ wordcloud2(data=word.freq.wof, size=1.2, minSize=3, color = "random-light", back 
word.stems <- unlist(hunspell_stem(all.words.wof))

word.freq.stems <- data.frame(table(word.stems))



wordcloud2(data=word.freq.wof, size=1.2, minSize=3, color = "random-light", backgroundColor = "black")

wc_stem <- wordcloud2(

  data = word.freq.stems,

  size = 1.2,

  minSize = 3,

  color = "random-light",

  backgroundColor = "black"

)



htmlwidgets::saveWidget(

  wc_stem,

  file = "03_report/wordcloud_stemmed.html",

  selfcontained = TRUE

)
 
 No newline at end of file

02_code/R/Game_Basic_Description.R

+5 −2
Original line number Diff line number Diff line 
# Clear workspace

rm(list = ls())



# Set working directory

setwd("C:/Users/golna.GOLNAZ/Desktop/Hohenheim/Data Science/Assignment")

# -----------------------------

# Setup

# -----------------------------

# Set working directory where data and scripts are stored



## setwd()





# Load the Datasets

02_code/R/task5_1_sentiment_bing_manual.R

+6 −2
Original line number Diff line number Diff line 
# Clear workspace

rm(list = ls())



# Set working directory

setwd("C:/Users/golna.GOLNAZ/Desktop/Hohenheim/Data Science/Assignment")

# -----------------------------

# Setup

# -----------------------------

# Set working directory where data and scripts are stored



## setwd()



load("01_data/reviews_final.RData")

02_code/R/task5_part2 & 3_ R package.R→02_code/R/task5_part2 & 3_syuzhet&nrc.R

+8 −3
Original line number Diff line number Diff line 
rm(list=ls())





# install.packages("syuzhet")

# Requires the 'syuzhet' package

library(syuzhet)



# Set working directory

setwd("C:/Users/golna.GOLNAZ/Desktop/Hohenheim/Data Science/Assignment")



# -----------------------------

# Setup

# -----------------------------

# Set working directory where data and scripts are stored



## setwd()



load("01_data/reviews_final_sentiment_manual.RData")

02_code/R/task5_part4_Sentiment_analysis_SharedDeepSeekData.R

deleted100644 → 0
+0 −188
Original line number Diff line number Diff line 
graphics.off()

rm(list = ls())



# =================================================

# Set working directory

# =================================================

setwd("C:/Users/golna.GOLNAZ/Desktop/Hohenheim/Data Science/Assignment")



# =================================================

# Load data (FULL DeepSeek)

# =================================================

load("01_data/sentiment_deepseek_255710.RData")

load("01_data/reviews_final_sentiment_manual.RData")

load("01_data/reviews_final_sentiment_syuzhet.RData")



# =================================================

# Extract LLM sentiment (CORRECT for this dataset)

# =================================================

llm_sentiment_all <- sentiment.ds$sentiment



# Sanity check

str(llm_sentiment_all)

summary(llm_sentiment_all)



# =================================================

# Figure 1: LLM sentiment distribution (ALL reviews)

# =================================================

sentiment_scores <- llm_sentiment_all



par(mar = c(5, 4, 4, 2) + 0.1)



hist(

  sentiment_scores,

  breaks = 30,

  freq = FALSE,

  col = "lightblue",

  border = "white",

  main = "Distribution of Sentiment Scores (All DeepSeek Reviews)",

  xlab = "Sentiment score (-1 = negative, +1 = positive)"

)



lines(

  density(sentiment_scores, na.rm = TRUE),

  lwd = 2

)



# =================================================

# Align data for comparisons

# =================================================

n_compare <- nrow(reviews_final)

llm_sentiment <- tail(llm_sentiment_all, n_compare)



# =================================================

# Figure 2: Manual Bing vs LLM

# =================================================

manual_sentiment <- reviews_final$sentiment_manual

keep <- is.finite(manual_sentiment) & is.finite(llm_sentiment)

sum(keep)

length(keep)



par(mar = c(5, 5, 4, 2) + 0.1)



plot(

  density(manual_sentiment[keep]),

  ylim = c(0, 4),

  col = "darkgray",

  lwd = 2,

  lty = 2,

  xlim = c(-1, 1),

  main = "Manual Bing vs LLM Sentiment",

  xlab = "Sentiment score",

  ylab = "Density"

)



lines(

  density(llm_sentiment[keep]),

  col = "black",

  lwd = 2

)



legend(

  "topleft",

  legend = c("Manual Bing", "LLM (DeepSeek)"),

  col = c("darkgray", "black"),

  lwd = 2,

  lty = c(2, 1),

  bty = "n"

)



# =================================================

# Figure 3: syuzhet 

# =================================================

syuzhet <- reviews_final$sentiment_syuzhet





par(mar = c(4, 4, 3, 1) + 0.1)



plot(

  density(syuzhet, na.rm = TRUE),

  col = "darkgray",

  lwd = 2,

  main = "syuzhet Bing Sentiment (Raw)",

  xlab = "Raw sentiment score",

  ylab = "Density"

)







# =================================================

# Figure 4: syuzhet vs LLM

# =================================================



par(mar = c(5, 5, 4, 2) + 0.1)



plot(

  density(syuzhet, na.rm = TRUE),

  ylim = c(0, 4),

  col = "darkgray",

  lwd = 2,

  lty = 2,

  xlim = c(-1, 1),

  main = "syuzhet vs LLM Sentiment",

  xlab = "Sentiment score (-1 to +1)",

  ylab = "Density"

)



lines(

  density(llm_sentiment),

  col = "black",

  lwd = 2

)



legend(

  "topleft",

  legend = c("syuzhet", "LLM (DeepSeek)"),

  col = c("darkgray", "black"),

  lwd = 2,

  lty = c(2, 1),

  bty = "n"

)



# =================================================

# Figure 5: NRC vs LLM

# =================================================





nrc_sentiment <- tail(

  reviews_final$sentiment_nrc,

  length(llm_sentiment)

)



keep <- is.finite(nrc_sentiment) & is.finite(llm_sentiment)



#pdf(file.path(output_path, "nrc_vs_llm_density.pdf"),

#    width = 7, height = 5)





par(mar = c(5, 5, 4, 2) + 0.1)



plot(

  density(nrc_sentiment[keep]),

  col = "darkgray",

  lwd = 2,

  lty = 2,

  xlim = c(-1, 1),

  ylim = c(0, 4),

  main = "NRC Sentiment vs LLM Sentiment",

  xlab = "Sentiment score",

  ylab = "Density"

)



lines(

  density(llm_sentiment[keep]),

  col = "black",

  lwd = 2,

  lty = 1

)



legend(

  "topleft",

  legend = c("NRC sentiment", "LLM (DeepSeek)"),

  col = c("darkgray", "black"),

  lwd = 2,

  lty = c(2, 1),

  bty = "n"

)



dev.off()