gistfile1.rb

#! /usr/bin/env ruby

require "json"

infile = "antarctica_hits"

lines = File.readlines infile
parsed_lines = lines.map do |line|
	line = line.gsub(/2014-02-28T12-ip-10.*\.internal:/, "")
	JSON.parse(line) rescue next
end.compact

puts "", "found #{lines.length} hit events"
puts "  WEB_INITIAL: #{parsed_lines.select {|h| h['eventType'] === "WEB_INITIAL"}.length}"
puts "  WEB_ADDITIONAL: #{parsed_lines.select {|h| h['eventType'] === "WEB_ADDITIONAL"}.length}", ""

hits = {}

parsed_lines.each do |hit|
	hits[hit['viewId']] ||= []

	if hit['eventType'] == "WEB_ADDITIONAL"
		begin
			hits[hit['viewId']] << hit['client']['page']['depth']
		rescue StandardError
		end
	end
end

hits.reject! {|k, v| v.empty?}
puts "averaging across #{hits.length} views (avg #{(hits.map {|k, d| d.length}.reduce(&:+) / hits.length.to_f).round(2)} events per view)"

depths = hits.map {|k, depths| depths.max}.flatten
total = depths.reduce(&:+)
mean = total.to_f / hits.length

puts "mean: #{mean.round(3)}%"