###############################################################################
# R Optimization Tutorial — Profiling with profvis
#
# Run each profvis({ ... }) block in RStudio. profvis opens an interactive
# view with two parts:
#   (1) a flame graph  — wide bars = where time is spent
#   (2) a data table   — line-by-line Time (ms) and Memory (MB)
#
# In the code panel of the profvis output, look for the longest memory and time bars.
#
###############################################################################

library(profvis)

## ---- Setup: build a sample dataset -----------------------------------------
set.seed(123)
n <- 100000                                   
df <- data.frame(
  id    = seq_len(n),
  group = sample(sprintf("g%03d", 1:200), n, replace = TRUE),  # 200 string groups
  value = rnorm(n, mean = 100, sd = 15)
)

###############################################################################
# SLOW VERSION 
###############################################################################
profvis({
  
  ## --- Bottleneck 1: growing a vector inside a loop (no pre-allocation) ------
  ## Every c() builds a brand-new, longer vector and copies all prior elements.
  doubled <- c()
  for (i in seq_len(n)) {
    doubled <- c(doubled, df$value[i] * 2)
  }
  
  ## --- Bottleneck 2: apply() over data-frame rows ---------------------------
  ## apply() coerces the WHOLE data frame to one matrix. Because `group` is a
  ## character column, the matrix becomes character, so every number is turned
  ## into a string and back. The R closure is also called once per row.
  df$normalized <- apply(df, 1, function(row) {
    as.numeric(row["value"]) / 100
  })
  
  ## --- Bottleneck 3: repeated linear-scan subsetting -------------------------
  ## df[df$group == g, ] rescans all 30,000 rows for each of the 200 groups
  ## and allocates a fresh subset every iteration.
  group_means <- c()
  for (g in sort(unique(df$group))) {
    sub <- df[df$group == g, ]
    group_means <- c(group_means, mean(sub$value))
  }
  
})