library(tidyverse)

neurobiber_csv <-"~/p2/quest/072325_biberplus_labels.csv"
neurobiber_df <- read.csv(neurobiber_csv , header = TRUE) 

normalized_cols <- grep("^normalized_", names(neurobiber_df), value = TRUE)

neurobiber_df$normalized_features_vec <- lapply(
  asplit(neurobiber_df[, normalized_cols], 1), as.numeric
)

X <- do.call(rbind, neurobiber_df$normalized_features_vec)


set.seed(808)

library(dplyr)
library(purrr)
table(neurobiber_df$source)

#neurobiber_df <- neurobiber_df %>%
#  group_by(source) %>%
#  mutate(cluster = {
#    X_sub <- do.call(rbind, features_vec)
#    as.factor(kmeans(X_sub, centers = 50)$cluster)
#  }) %>%
#  ungroup()
library(dbscan)
dbscan_result <- dbscan(X, eps = 0.5, minPts = 97)
neurobiber_df$dbcluster <- as.factor(ifelse(dbscan_result$cluster == -1, "Noise", dbscan_result$cluster))

kmeans_result <- kmeans(X, centers = 10)
neurobiber_df$cluster <- as.factor(kmeans_result$cluster)
table(neurobiber_df$dbcluster)

pca <- prcomp(X, center = TRUE, scale. = TRUE)
neurobiber_df$PC1 <- pca$x[,1]
neurobiber_df$PC2 <- pca$x[,2]


ggplot(neurobiber_df, aes(x = PC1, y = PC2, color = phase)) +
  geom_point(size = 2, alpha = 0.7) +
  theme_minimal() +
  labs(title = "Across-case comment clusters (DBSCAN) by cross-case PCA",
       x = "Principal Component 1",
       y = "Principal Component 2") +
  facet_wrap(~ source)

ggplot(neurobiber_df, aes(x = phase, y=dbcluster, fill=AuthorWMFAffil)) +
  geom_violin(trim = FALSE, position = position_dodge(width = 0.8), alpha = 0.6) +
  theme_minimal() +
  labs(title = "Across-case comment clusters by feature deployment phase",
       x = "Feature deployment phase",
       y = "Neurobiber feature vector cluster (DBSCAN)") +
  facet_wrap(~ source)

cluster_means <- aggregate(
  X,
  by = list(
    WMFAffil = neurobiber_df$AuthorWMFAffil,
    phase = neurobiber_df$phase,
    comment_type = neurobiber_df$comment_type,
    source= neurobiber_df$source
  ),
  FUN = mean
)

rownames(cluster_means) <- apply(
  cluster_means[, c("WMFAffil", "phase", "comment_type", "source")], 1,
  function(x) paste(x, collapse = "_")
)

cluster_means <- cluster_means[, !(names(cluster_means) %in% c("WMFAffil", "phase", "comment_type", "source"))]

#cluster_means <- aggregate(X, by = list(Cluster = neurobiber_df$AuthorWMFAffil), FUN = mean)
#rownames(cluster_means) <- paste0("Cluster_", cluster_means$Cluster)
#cluster_means <- cluster_means[,-1] # Remove cluster label column

BIBER_FEATURES <- c(
  "BIN_QUAN","BIN_QUPR","BIN_AMP","BIN_PASS","BIN_XX0","BIN_JJ",
  "BIN_BEMA","BIN_CAUS","BIN_CONC","BIN_COND","BIN_CONJ","BIN_CONT",
  "BIN_DPAR","BIN_DWNT","BIN_EX","BIN_FPP1","BIN_GER","BIN_RB",
  "BIN_PIN","BIN_INPR","BIN_TO","BIN_NEMD","BIN_OSUB","BIN_PASTP",
  "BIN_VBD","BIN_PHC","BIN_PIRE","BIN_PLACE","BIN_POMD","BIN_PRMD",
  "BIN_WZPRES","BIN_VPRT","BIN_PRIV","BIN_PIT","BIN_PUBV","BIN_SPP2",
  "BIN_SMP","BIN_SERE","BIN_STPR","BIN_SUAV","BIN_SYNE","BIN_TPP3",
  "BIN_TIME","BIN_NOMZ","BIN_BYPA","BIN_PRED","BIN_TOBJ","BIN_TSUB",
  "BIN_THVC","BIN_NN","BIN_DEMP","BIN_DEMO","BIN_WHQU","BIN_EMPH",
  "BIN_HDG","BIN_WZPAST","BIN_THAC","BIN_PEAS","BIN_ANDC","BIN_PRESP",
  "BIN_PROD","BIN_SPAU","BIN_SPIN","BIN_THATD","BIN_WHOBJ","BIN_WHSUB",
  "BIN_WHCL","BIN_ART","BIN_AUXB","BIN_CAP","BIN_SCONJ","BIN_CCONJ",
  "BIN_DET","BIN_EMOJ","BIN_EMOT","BIN_EXCL","BIN_HASH","BIN_INF",
  "BIN_UH","BIN_NUM","BIN_LAUGH","BIN_PRP","BIN_PREP","BIN_NNP",
  "BIN_QUES","BIN_QUOT","BIN_AT","BIN_SBJP","BIN_URL","BIN_WH",
  "BIN_INDA","BIN_ACCU","BIN_PGAS","BIN_CMADJ","BIN_SPADJ","BIN_X"
)
BIBER_FEATURES_NO_BIN <- gsub("^BIN_", "", BIBER_FEATURES)
colnames(cluster_means) <- BIBER_FEATURES_NO_BIN
library(pheatmap)
pheatmap(cluster_means, 
         cluster_rows = FALSE,    # Now features are clustered (rows)
         cluster_cols = FALSE,
         scale='none')         # Standardize features