happiday.R

## fetch & parse data from Wikipedia
library(XML)
wiki <- 'https://en.wikipedia.org/wiki/Chronology_of_computation_of_%CF%80'
tables <- readHTMLTable(readLines(wiki), stringsAsFactors = FALSE)

## merge data from 4 tables
library(data.table)
pis <- rbindlist(list(

    ## data data before 1400 from 3rd table extracted by hand
    fread('Date,Whom,Decimal places(world records in bold)
-250,Archimedes, 2
150,Ptolemy,3
263,Liu Hui,4
480,Zu Chongzhi,6', sep = ',', header = TRUE),

    ## no computes
    tables[[4]][, c(1, 2, 4)],

    ## mainframes
    tables[[5]][, c(1, 2, 5)],

    ## PCs
    tables[[6]][, c(1, 2, 5)]))

## clean column names
setnames(pis, c('date', 'name', 'digits'))

## remove rows without valid digits
pis[, digits := as.numeric(gsub(',', '', digits))]
pis <- pis[!is.na(digits)]

## drop non record rows
pis <- pis[digits - cummax(digits) == 0]

## clean references from names
pis[, name := sub('\\[.*\\]', '', name)]

## keep only first person's name
pis[, name := sub('(.*)( (and|&|,)).*', '\\1', name)]

## clean up dates
pis[date == 'Late 18th century', date := '1799'] # this is pretty late
pis[, date := as.integer(sub('.* ([0-9]*)$', '\\1', date))]

## ggplot + repel + emoGG magic
library(ggplot2)
library(ggrepel); library(grid)
library(emoGG)
ggplot(pis, aes(date, digits)) + geom_step() +
    geom_emoji(emoji = '1f370', size = sqrt(pi) / 1e2) +
    geom_text_repel(aes(label = sprintf('%s (%s)', name, date)),
        size = pi, max.iter = pi * 1e3) +
    scale_y_log10(breaks = 10^(0:13)) +
    scale_x_continuous(breaks = c(-250, 0, 150, 263, 480, 1400, 1600, 1800, 1900, 2000)) +
    theme_bw() + xlab('') + ylab('Decimal places of π') +
        ggtitle(expression(atop(
            'Chronology of computation of π',
            atop(italic('Based on data fetched from Wikipedia')))))

happiday2016.png