module Scotchit

Constants

DB: seeding the hashes
Price
Stats
VERSION

Public Class Methods

ci_lower_bound(pos, n, confidence) click to toggle source

In order with the philosophy of reddit, this program evaluates the ratings that redditors have given to all of the whiskys submitted to review by executing a Wilson Score confidence interval (lower bound). This is the same algroithm that reddit users interally for ratings and thus what you see when you visit the site.

Evan Miller's Wilson Score Interval fn

# File lib/scotchit.rb, line 33
def ci_lower_bound(pos, n, confidence)
    if n == 0
        return 0
    end
    z = Statistics2.pnormaldist(1-(1-confidence)/2)
    phat = 1.0*pos/n
    (phat + z*z/(2*n) - z * Math.sqrt((phat*(1-phat)+z*z/(4*n))/n))/(1+z*z/n)
end

parse_csv(file_name) click to toggle source

grok the ratings.csv file

# File lib/scotchit.rb, line 44
def parse_csv(file_name)
    CSV.foreach(file_name, headers:true) do |row|
        # ignore non-Scotch things
        if %w[Bourbon Rye Grain Tennessee Liqueur Wheat].include? row['Region']
            next
        end
        # clean up key name
        name = row["Whisky Name"].strip
        # get the rating as an integer
        val = row['Rating'].to_i
        # get the cost as a floating point by removing non-digits and throwing out
        # un-american funny money
        if !%w[£ CDN CAD AUD € GBP NZD EUR CAN].any? {|x| row['Price'].to_s.upcase.include? x} #ugly
            cost = row['Price'].to_s.gsub(/[^\d\.]/, '').to_f
        else
            cost = 0.0
        end
        # seed key:val
        if !DB.has_key?(name)
            DB[name] = []
            Price[name] = []
        end
        # append score
        DB[name] << ((val == nil) ? 0 : val)
        if cost > 20.0
            Price[name] << cost
        end
    end
end

run() click to toggle source

reverse sort by confidence score and print to term

we are only displaying whole number percentages in order to avoid the presception that this is accurate to some decimal of score. we're trying to select scotches that have good confidence of being yummy and deserving of the price via the sample of redditors.

# File lib/scotchit.rb, line 102
def run()
    parse_csv("lib/ratings.csv")
    score()
    Stats.sort_by {|k,i| -i[0]}.each {|k,i| (i[0] > 20.0 && (puts "#{k}: #{i[0]}% (#{i[1]}+, #{i[2]}#) #{i[3]}")) || nil}
end

score() click to toggle source

do the confidence interval computations for all scotches meeting our thresholds.

# File lib/scotchit.rb, line 76
def score()
    DB.keys.each do |k|
        if DB[k].count > 5 # only eval whisky that has 6 or more reviews on reddit
            pos = DB[k].count {|x| x > 90 } # threshold as positive vote 91+% rating
            # calculate the Price indicator
            average = (Price[k] != []) ? Price[k].reduce(:+) / Price[k].count : 0.0
            indicator = case average.round(0)
            when 0 then "?"
            when 1..39 then "$"
            when 40..69 then "$$"
            when 70..89 then "$$$"
            when 90..120 then "$$$$"
            else
                "$$$$$"
            end
            Stats[k] = [(ci_lower_bound(pos, DB[k].count, 0.975) * 100.0).round(0), pos, DB[k].count, indicator]
        end
    end
end

Private Instance Methods