escher/data/transcript_freq_study.R

library(foreach)
library(iterators)
library(ggplot2)
library(dplyr)

#------
messages <- read.csv2("transcript/minecraft_data-1659808309634.csv", sep = ",")
messages <- messages %>% mutate(
  start_time = as.POSIXct(strptime(start_time, "%F %T")),
  end_time = as.POSIXct(strptime(end_time, "%F %T"))
)

n_m <- nrow(messages)

#--------

ggplot(messages, aes(x=start_time)) + theme_minimal() + geom_histogram(binwidth = 3600)
ggplot(messages %>% transmute(tdiff = difftime(start_time, lag(start_time), units = "secs")), aes(x=tdiff)) + theme_minimal() + 
  stat_ecdf() +
  stat_ecdf(data=data.frame(tdiff=rlomax(n_m*10, 1/(b_p + len),
                                         (a_p + n_m))), color="red")

ggplot(messages %>% transmute(tdiff = as.numeric(difftime(start_time, lag(start_time)), units = "secs")), aes(x=tdiff)) + theme_minimal() +
  geom_histogram(binwidth = 10) +
  geom_vline(xintercept = qdiff)

ggplot(messages, aes(x=start_time)) + theme_minimal() + geom_vline(aes(xintercept=start_time)) + geom_histogram(binwidth = 3600)

ggplot(messages %>% mutate(tdiff = difftime(start_time, lag(start_time), units = "secs")), aes(x=start_time, y=tdiff)) + theme_minimal() +
  geom_point() +
  geom_hline(yintercept = qdiff)

#-------

# Print messages

for(i in 1:nrow(messages)) {
  line = messages[i,]
  
  cat(sprintf("[%s] <%s> %s\n", line$start_time, line$names, line$transcription_line))  
  
  if(difftime(messages[i+1,"start_time"], line$start_time, units = "secs") > qdiff + 20)
    cat("----------------------\n")
}

# Show where messages were selected
ggplot(messages, aes(x=pub_time)) + theme_minimal() + 
  geom_histogram(binwidth = 600) + 
  geom_histogram(data=fmess, binwidth = 600, fill="green", alpha=0.7)

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

# Bayesian
# p(d < 3/60) = 0.01
# d ~ exp(l)
# l ~ gamma(a_p, b_p)
# l = argmax_l dgamma(a_p, 60)

a_p <- 13
b_p <- 60

# Total length of messages
# len <- as.numeric(difftime(messages[n_m,]$pub_time, messages[1,]$pub_time, units = "secs"))
len <- 697666.923
# Poisson lambda for messages per second

library(extraDistr)

gen_diffs <- rlomax(
  n_m, 
  1/(b_p + len),
  (a_p + n_m)
)

ggplot() + geom_histogram(mapping = aes(x=cumsum(gen_diffs)), binwidth = 3600)

# Plot with overlaid posterior
ggplot(messages, aes(x=start_time)) + theme_minimal() + 
  geom_histogram(binwidth = 600) + 
  geom_histogram(mapping = aes(x=cumsum(rexp(n_m, lambda)) + messages[1,"start_time"]), binwidth = 600, fill="red", alpha=0.3)

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

# Diff with 0.01 quantile
qdiff <- qlomax(0.02,   1/(b_p + len), (a_p + n_m))

fmess <- messages %>% filter(
  as.numeric(difftime(start_time, lag(start_time), units = "secs")) < qdiff |
    as.numeric(difftime(lead(start_time), start_time, units = "secs")) < qdiff
)

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

ggplot(messages %>% mutate(len = as.double(difftime(end_time, start_time, unit="secs"))), aes(x=len)) + geom_histogram(binwidth = 0.5)

ggplot(messages[1:50,], aes(xmin=start_time, xmax=end_time, ymin=-1, ymax = 1)) + theme_minimal() + geom_rect()

#---------

last <- 1
for (i in 2:nrow(messages)) {
  if (messages[i,]$start_time == messages[last,]$end_time) {
    messages[last,]$transcription_line <- paste(messages[last,]$transcription_line, messages[i,]$transcription_line)
    messages[last,]$end_time <- messages[i,]$end_time
    messages[i,]$transcription_line <- NA
  } else {
    last <- i
  }
}
messages <- messages %>% filter(!is.na(transcription_line))
messages %>% transmute(touch = start_time == lag(end_time))