Update analysis scripts and results (c2306d11) · Commits · Amir Raoufi / PoDS_Ass_Team31

00_docs/Schedule.pdf

deleted100644 → 0

−77.8 KiB

File deleted.

View file

01_data/reviews_final_sentiment_syuzhet.RData

+5.32 KiB (635 KiB)

File changed.

No diff preview for this file type.

View original file

View changed file

02_code/R/task5_part2 & 3_syuzhet&nrc.R

+4 −1

Original line number	Diff line number	Diff line
		@@ -76,7 +76,10 @@ summary(reviews_final$sentiment_nrc)
		# Overall, the game is perceived positively, but the scores vary because
		# NRC counts emotion words and depends on review length.


		#-------------------------------------------
		# Add a column containing the normalized nrc
		#-------------------------------------------
		reviews_final$sentiment_nrc_tanh <- tanh(reviews_final$sentiment_nrc)

		# save the file
		save(reviews_final,

02_code/R/task5_part4_sentiment_analysis_llm_nrc.R

+120 −24

Original line number	Diff line number	Diff line
		@@ -7,7 +7,7 @@ graphics.off()
		# ---------------------------
		# Set working directory
		# ---------------------------
		## setwd()
		#setwd()

		# ---------------------------------------------------------------------------
		# Load data
		@@ -104,6 +104,7 @@ barplot(
		ylim = c(0, 5000)
		)


		# Most frequent NRC emotion
		sum(reviews_merged$nrc_emotion_dominant == "anticipation", na.rm = TRUE)

		@@ -123,6 +124,10 @@ reviews_merged$emotion_LLM <- vapply(

		llm_counts <- table(reviews_merged$emotion_LLM, useNA = "no")


		plot_llm_emotions <- function() {
		par(mar = c(8, 6, 4, 2))

		barplot(
		llm_counts,
		main = "Emotions (LLM)",
		@@ -133,16 +138,15 @@ barplot(
		ylim = c(0, 5000)
		)

		mtext("Emotion", side = 1, line = 6)
		mtext("Frequency", side = 2, line = 5)

		par(mar = c(5, 4, 4, 2))
		}


		plot_llm_emotions()

		barplot(
		llm_counts,
		main = "Emotions (LLM)",
		xlab = "Emotion",
		ylab = "Frequency",
		col = "lightgreen",
		las = 2,
		ylim = c(0, 5000)
		)

		# Most frequent LLM emotion
		sum(reviews_merged$emotion_LLM == "joy", na.rm = TRUE)
		@@ -190,6 +194,7 @@ points(
		pch = 19
		)


		# ---------------------------------------------------------------------------
		# 6) Extreme sentiment and helpfulness
		# ---------------------------------------------------------------------------
		@@ -273,7 +278,6 @@ library(ggplot2)
		library(dplyr)
		library(tidyr)

		# Density comparison (same colors and format)
		df_long <- data.frame(
		score = c(
		reviews_final$sentiment_manual,
		@@ -286,7 +290,7 @@ df_long <- data.frame(
		)
		)

		ggplot(df_long, aes(x = score, color = method)) +
		p_density <- ggplot(df_long, aes(x = score, color = method)) +
		geom_density(size = 1.2) +
		scale_color_manual(values = c(
		"Manual" = "lightgreen",
		@@ -302,6 +306,18 @@ ggplot(df_long, aes(x = score, color = method)) +
		theme_minimal() +
		theme(legend.position = "right")

		# show the plot
		print(p_density)

		# save the same plot
		ggsave(
		filename = "03_report/graphs/density_sentiment_comparison.png",
		plot = p_density,
		width = 7,
		height = 5,
		dpi = 300
		)

		# ---------------------------------------------------------------------------
		# Barplot: Emotion distribution comparison (LLM vs NRC)
		# ---------------------------------------------------------------------------
		@@ -319,7 +335,7 @@ emotion_df <- rbind(
		)
		)

		ggplot(emotion_df, aes(x = reorder(emotion, count, FUN = mean),
		p_emotion <- ggplot(emotion_df, aes(x = reorder(emotion, count, FUN = mean),
		y = count,
		fill = method)) +
		geom_col(position = "dodge", alpha = 0.8) +
		@@ -335,4 +351,84 @@ ggplot(emotion_df, aes(x = reorder(emotion, count, FUN = mean),
		theme_minimal() +
		theme(legend.position = "right")

		# show
		print(p_emotion)

		# save
		ggsave(
		"03_report/graphs/emotion_comparison.png",
		plot = p_emotion,
		width = 7,
		height = 5,
		dpi = 300
		)


		# ---------------------------------------------------------------------------
		# Prepare data for boxplot (LLM vs NRC by voted_up)
		# ---------------------------------------------------------------------------

		df_box <- bind_rows(

		# LLM sentiment (already in [-1, 1])
		reviews_merged %>%
		transmute(
		voted_up,
		sentiment_score = sentiment,
		method = "LLM"
		),

		# NRC sentiment (tanh-normalized to [-1, 1])
		reviews_merged %>%
		transmute(
		voted_up,
		sentiment_score = sentiment_nrc_tanh,
		method = "NRC"
		)
		)

		# ---------------------------------------------------------------------------
		# Boxplot
		# ---------------------------------------------------------------------------

		p_box <- ggplot(
		df_box,
		aes(x = voted_up, y = sentiment_score, fill = method)
		) +
		geom_boxplot(
		alpha = 0.7,
		position = position_dodge(width = 0.8)
		) +
		stat_summary(
		fun = mean,
		geom = "point",
		shape = 19,
		size = 3,
		color = "red",
		position = position_dodge(width = 0.8)
		) +
		scale_fill_manual(
		values = c("LLM" = "lightblue", "NRC" = "lightcoral")
		) +
		labs(
		x = "Voted Up",
		y = "Sentiment score (-1 to 1)",
		fill = "Method"
		) +
		theme_minimal() +
		theme(legend.position = "right")

		# show plot
		print(p_box)

		ggsave(
		"03_report/graphs/sentiment_boxplot_LLM_NRC.png",
		plot = p_box,
		width = 7,
		height = 5,
		dpi = 300
		)

03_report/graphs/Log-transformed total playtime.pdf

deleted100644 → 0

−4.82 KiB

File deleted.

View file