# 70, 0.13 denotes that about 13% domains have at least 70 snapshots
# Probable admin domains criterias:
# Total domains: ~870K domains
# 1. - Part of alexa: ~23K domains
# 2. - Having less than 5 snapshots: ~600K domains
# 3. - Domains having query on a 48 hr frequency: ~150K domains

reset
set xlabel "No. of snapshots"
set ylabel "CDF"
set datafile separator ","
# set title "CDF vs Hours"
set terminal pngcairo
set output 'figures/cdf-vs-no-of-analyses.png'
set key bottom

set xrange [1:29064]
set yrange [0:1]
# set xtics (1,10,50,200,500,1000,5000,10000,30000)
set logscale x 2

plot "../data/cdf-vs-no-of-snapshots.txt" using 1:2 w lines lc rgb 'dark-blue' dt 2 title 'Alexa domains', \
"../data/cdf-vs-no-of-snapshots.txt" using 1:3 w lines lc rgb 'brown' title 'Non-alexa domains', \
"../data/cdf-vs-no-of-snapshots.txt" using 1:4 w lines lc rgb 'black' title 'Probable admin domains'