library(tidyverse)

neurobiber_csv <-"~/p2/quest/072525_pp_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
)
library(dplyr)
# duplicate, declined, invalid -> declined 
# stalled, open, progress -> open
# resolved -> resolved
neurobiber_df <- neurobiber_df |>
  filter(comment_type == "task_description") |>
  mutate(
    task_status = case_when(
      status %in% c("duplicate", "declined", "invalid") ~ "declined",
      status %in% c("stalled", "open", "progress") ~ "open",
      status == "resolved" ~ "resolved",
      TRUE ~ status  # fallback for unexpected values
    ))

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

library(coop)
#cos_sim1 <- coop::cosine(t(X))

register_means <- aggregate(
  X,
  by = list(
    priority = neurobiber_df$priority,
    outcome= neurobiber_df$task_status,
    phase = neurobiber_df$phase,
    source = neurobiber_df$source,
    affiliation = neurobiber_df$AuthorWMFAffil
  ),
  FUN = mean
)

feature_mat <- as.matrix(register_means[, -(1:5)])  
cos_sim_matrix <- coop::cosine(t(feature_mat))
rownames(cos_sim_matrix) <- apply(register_means[, 1:5], 1, paste, collapse = "_")
colnames(cos_sim_matrix) <- rownames(cos_sim_matrix)


scaled_mat <- scale(cos_sim_matrix)
#pheatmap(scaled_mat, symm = TRUE)
#heatmap(cos_sim_matrix, col=heat.colors(256), breaks=seq(-1, 1, length.out=257))
library(viridis)
library(pheatmap)
pheatmap(cos_sim_matrix, 
         cluster_rows = FALSE,    # Now features are clustered (rows)
         cluster_cols = FALSE,
         scale='none',
         color = viridis(100))         # Standardize featu

diag(cos_sim_matrix) <- NA
which(cos_sim_matrix == max(cos_sim_matrix, na.rm = TRUE), arr.ind = TRUE)  # Most similar
which(cos_sim_matrix == min(cos_sim_matrix, na.rm = TRUE), arr.ind = TRUE)  # Least similar