tidy_report.R in R

A whole tidyverse pipeline: read, clean, reshape, model per group, plot.

suppressPackageStartupMessages({
  library(dplyr)
  library(tidyr)
  library(readr)
  library(purrr)
  library(stringr)
  library(ggplot2)
})

RAW <- "session,lang,chars,seconds,errors
1,r,1520,60,4
2,r,1680,60,3
3,r,1810,60,2
4,r,1900,60,2
1,python,1740,60,5
2,python,1880,60,3
3,python,2010,60,1
4,python,2120,60,1
1,sql,1180,60,6
2,sql,1240,60,4
3,sql,1310,60,3
"

runs <- read_csv(RAW, show_col_types = FALSE) |>
  mutate(
    wpm = (chars / 5) / (seconds / 60),
    accuracy = (1 - errors / chars) * 100,
    lang = str_to_title(lang)
  )

cat("--- per language ---\n")
summary_table <- runs |>
  summarise(
    sessions = n(),
    best_wpm = max(wpm),
    mean_wpm = mean(wpm),
    mean_accuracy = mean(accuracy),
    .by = lang
  ) |>
  arrange(desc(mean_wpm)) |>
  mutate(across(where(is.numeric), \(x) round(x, 1)))
print(summary_table)

cat("\n--- improvement per session ---\n")
slopes <- runs |>
  nest(.by = lang) |>
  mutate(
    model = map(data, \(d) lm(wpm ~ session, data = d)),
    per_session = map_dbl(model, \(m) coef(m)[["session"]]),
    r_squared = map_dbl(model, \(m) summary(m)$r.squared)
  ) |>
  select(lang, per_session, r_squared) |>
  mutate(across(where(is.numeric), \(x) round(x, 3)))
print(slopes)

cat("\n--- wide view ---\n")
print(runs |>
        select(lang, session, wpm) |>
        mutate(wpm = round(wpm)) |>
        pivot_wider(names_from = session, values_from = wpm,
                    names_prefix = "s"))

plot <- runs |>
  ggplot(aes(session, wpm, colour = lang)) +
  geom_point(size = 2) +
  geom_smooth(method = "lm", formula = y ~ x, se = FALSE, linewidth = 0.8) +
  scale_x_continuous(breaks = 1:4) +
  labs(title = "Words per minute by session", x = "session", y = "wpm") +
  theme_minimal()

out <- tempfile(fileext = ".png")
ggsave(out, plot, width = 6, height = 3.5, dpi = 120)
cat("\nwrote", basename(out), "\n")

How it works

1. Every stage is a verb in one pipeline, with no stray variables.
2. The per-group model lives in a list column.
3. The plot comes off the end of the same pipeline.

Keywords and builtins used here

• across
• aes
• arrange
• basename
• cat
• coef
• desc
• geom_point
• geom_smooth
• ggplot
• ggsave
• labs
• library
• lm
• map
• map_dbl
• max
• mean
• mutate
• n
• nest
• pivot_wider
• print
• read_csv
• round
• scale_x_continuous
• select
• str_to_title
• summarise
• summary
• suppressPackageStartupMessages
• tempfile
• theme_minimal
• where

The run, in numbers

Lines

73

Characters to type

1736

Tokens

441

Three-star pace

110 tpm

At the three-star pace of 110 tokens a minute, this run takes about 241 seconds.

Type this snippet

Step 1 of 1 in Encore, step 12 of 12 in Reshaping & reading.

← Previous