new alg test

newpopalg.rb

require 'date'
require 'csv'

#Given a csv file output from Sermo App at https://app.sermo.com/reporting/reports/post_summary_by_date
#we need a csv of the posts in order of popularity, with the addition of a bayesian average to respond to different days.

#constructor of my post object with the csv fields to hold all the attributes
class Post <

  # a post has csv categories a-q
  Struct.new(:post_id, :title, :link, :categories, :publish_date, :avg_rating, :raters, :comments, :voters, :votes, :views, :author_id, :username, :first_name, :last_name, :email, :specialty, :pop)

end

# define an array to hold the all the posts of the week
posts = Array.new

#asks me what the file name for the week is and removes enter
csv = gets.chomp

# open csv file and then grab each field, skipping top row of field names
CSV.foreach(csv) {|fields|
  if fields[0] == 'Post ID' then
    next
  end
  
  # create a new Post constructor object
  p = Post.new
  #methods after field object to get rid of double-quotes and blanks
  p.post_id = fields[0].tr_s('"', '').strip
  p.title = fields[1].tr_s('"', '').strip
  p.link = fields[2].tr_s('"', '').strip
  p.categories = fields[3].tr_s('"', '').strip
  #taking the hours and minutes off the date field
  date = fields[4].split(' ')[0]
  #then convert that date variable to the ruby Date object
  p.publish_date =  Date.parse(date)
  p.avg_rating = fields[5].tr_s('"', '').strip.to_f
  p.raters = fields[6].tr_s('"', '').strip.to_i
  p.comments = fields[7].tr_s('"', '').strip.to_i
  p.voters = fields[8].tr_s('"', '').strip.to_i
  p.votes = fields[9].tr_s('"', '').strip.to_i
  p.views = fields[10].tr_s('"', '').strip.to_i
  p.author_id = fields[11].tr_s('"', '').strip
  p.username = fields[12].tr_s('"', '').strip
  p.first_name = fields[13].tr_s('"', '').strip
  p.last_name = fields[14].tr_s('"', '').strip
  p.email = fields[15].tr_s('"', '').strip
  #for non-physician accounts a nil value of specialty appears which this skips
  if fields[16].nil? then
    next
  end
  p.specialty = fields[16].tr_s('"', '').strip
  #puts the items p into the array posts
  posts.push(p)
}

#need a bayesian constant to adjust against five star bandit/zero star bandit
#bayesian constant will be the average number raters on a post, or avg_num_raters
#for the rest of the pop formula: ( (avg_num_raters * avg_rating) + (this_num_raters * this_rating) ) / (avg_num_raters + this_num_raters)
#we need to get the average rating, average comments, average views, and average voters

#average number of ratings
avg_num_raters = posts.inject(0.0){|sum, p| sum + p.raters} / posts.size
ar = avg_num_raters
puts ar

avg_rating = posts.inject(0.0){|sum, p| sum + p.avg_rating} / posts.size
s = avg_rating
puts s

#average number of voters
avg_voters = posts.inject(0.0){|sum, p| sum + p.voters} / posts.size
o = avg_voters
puts o

#average number of views
avg_views = posts.inject(0.0){|sum, p| sum + p.views} / posts.size
v = avg_views
puts v


popped = Array.new
posts.each {|p|
  #applies new popularity alg to each post
  p.pop = p.comments + ( (ar * s) + (p.raters * p.avg_rating) ) / (o + p.voters) + (v*p.views)
     popped.push(p)
}

#outputing a csv file with today's date
daily = File.open("Day of " + posts.last.publish_date.next.strftime("%m%d%y") + ".csv", "w")

#adding new field
  popped.each {|p|
    row = [p.pop, p.publish_date, p.post_id, p.author_id, p.username, p.first_name, p.last_name, p.email]
    popped.puts row.join(",")
    }
    popped.close